#include "lemon/server.h"
#include "lemon/audio_types.h"
#include "lemon/auto_tune.h"
#include "lemon/error_types.h"
#include <optional>
#include "lemon/auto_tune.h"
#include "lemon/collection_orchestrator.h"
#include "lemon/hf_variants.h"
#include "lemon/model_registry.h"
#include "lemon/route_decision_response.h"
#include "lemon/routing_classifier_services.h"
#include "lemon/routing_policy.h"
#include "lemon/routing_policy_parser.h"
#include "lemon/config_file.h"
#include "lemon/jobs/job_manager.h"
#include "lemon/mcp_server.h"
#include "lemon/mcp_client.h"
#include "lemon/ollama_api.h"
#include "lemon/backends/backend_descriptor_registry.h"
#include "lemon/backends/cloud/cloud_server.h"
#include "lemon/backends/sdcpp/sdcpp_server.h"
#include "lemon/backends/thenoise/thenoise_server.h"
#include "lemon/backends/backend_utils.h"
#include <cstring>
#include "lemon/utils/conversation_fingerprint.h"
#include "lemon/utils/image_sniff.h"
#include "lemon/utils/json_utils.h"
#include "lemon/utils/model_name_utils.h"
#include "lemon/utils/path_utils.h"
#include "lemon/streaming_proxy.h"
#include "lemon/logging_config.h"
#include "lemon/thinking_controls.h"
#include "lemon/prometheus_metrics.h"
#include "lemon/runtime_config.h"
#include "telemetry.h"
#include "lemon/system_info.h"
#include "lemon/version.h"
#include <cctype>
#include <cstdint>
#include <cstdlib>
#include <iostream>
#include <iomanip>
#include <sstream>
#include <fstream>
#include <map>
#include <memory>
#include <thread>
#include <chrono>
#include <mutex>
#include <filesystem>
#include <system_error>
#include <algorithm>
#include <cmath>
#include <set>
#include <vector>
#include <lemon/utils/aixlog.hpp>
#include "lemon/utils/network_utils.h"
#include "lemon/utils/origin_utils.h"

#ifdef _WIN32
    #include <windows.h>
    #include <winsock2.h>
    #include <ws2tcpip.h>
#else
    #include <arpa/inet.h>   // inet_pton, htons
    #include <fcntl.h>
    #include <netdb.h>       // Crucial for getaddrinfo and addrinfo struct
    #include <netinet/in.h>  // sockaddr_in / sockaddr_in6
    #include <sys/socket.h>
    #include <sys/types.h>
    #include <unistd.h>
#endif

#ifdef __APPLE__
    #include <sys/sysctl.h>
#endif

#ifdef __linux__
    #include <sys/ioctl.h>
    #include <fcntl.h>
    #include <unistd.h>
    #include <libdrm/drm.h>
    #include "lemon/amdxdna_accel.h"
#endif

namespace fs = std::filesystem;

namespace lemon {

namespace {



// Normalize client-provided model names: strip ":latest" suffix (Ollama/Docker convention)
// Returns true if the model name was modified
bool normalize_client_model_name(json& request_json) {
    if (!request_json.contains("model") || !request_json["model"].is_string()) {
        return false;
    }

    std::string model_name = request_json["model"].get<std::string>();
    const std::string latest_suffix = ":latest";

    if (model_name.size() > latest_suffix.size() &&
        model_name.substr(model_name.size() - latest_suffix.size()) == latest_suffix) {
        std::string normalized = model_name.substr(0, model_name.size() - latest_suffix.size());
        request_json["model"] = normalized;
        return true;
    }

    return false;
}


bool valid_error_status(int status_code) {
    return status_code >= 400 && status_code <= 599;
}

int get_error_status_code(const json& response, int default_status_code = 500) {
    if (!response.contains("error") || !response["error"].is_object()) {
        return default_status_code;
    }

    const auto& error = response["error"];
    if (error.contains("status_code") && error["status_code"].is_number_integer()) {
        int status_code = error["status_code"].get<int>();
        if (valid_error_status(status_code)) {
            return status_code;
        }
    }

    if (error.contains("details") && error["details"].is_object()) {
        const auto& details = error["details"];
        if (details.contains("status_code") && details["status_code"].is_number_integer()) {
            int status_code = details["status_code"].get<int>();
            if (valid_error_status(status_code)) {
                return status_code;
            }
        }
    }

    if (error.contains("type") && error["type"].is_string()) {
        const std::string type = error["type"].get<std::string>();

        if (type == ErrorType::INVALID_REQUEST ||
            type == "invalid_request_error" ||
            type == ErrorType::UNSUPPORTED_OPERATION) {
            return 400;
        }

        if (type == ErrorType::MODEL_NOT_LOADED) {
            return 404;
        }
    }

    return default_status_code;
}

void set_error_response(const json& response, httplib::Response& res,
                        int default_status_code = 500) {
    res.status = get_error_status_code(response, default_status_code);
    res.set_content(response.dump(), "application/json");
}

void set_router_residency_conflict_response(
    const RouterResidencyConflictException& error,
    httplib::Response& res) {
    res.status = 409;
    const json response = {
        {"error", {
            {"message", error.what()},
            {"type", ErrorType::ROUTER_RESIDENCY_CONFLICT},
            {"param", "model"},
            {"code", ErrorType::ROUTER_RESIDENCY_CONFLICT},
        }},
    };
    res.set_content(response.dump(), "application/json");
}

void attach_route_decision(json& response, httplib::Response& res,
                           const std::optional<RouterDispatchResult>& dispatch) {
    if (!dispatch.has_value()) {
        return;
    }
    response["x_lemonade_route"] = route_decision_to_json(dispatch->decision);
    attach_route_header(res, dispatch->decision);
}

template <typename StreamFn>
void set_route_decision_sse_content_provider(
    httplib::Response& res,
    const std::optional<RouterDispatchResult>& dispatch,
    std::string request_body,
    StreamFn stream_fn) {
    res.set_header("Cache-Control", "no-cache");
    res.set_header("Connection", "keep-alive");
    res.set_header("X-Accel-Buffering", "no");
    if (dispatch) {
        attach_route_header(res, dispatch->decision);
    }

    json route_decision_json = dispatch
        ? route_decision_to_json(dispatch->decision)
        : json(nullptr);
    res.set_chunked_content_provider(
        "text/event-stream",
        [request_body = std::move(request_body),
         route_decision_json = std::move(route_decision_json),
         stream_fn = std::move(stream_fn)](size_t offset, httplib::DataSink& sink) {
            if (offset > 0) {
                return false;
            }

            stream_with_route_decision(
                sink,
                route_decision_json,
                [&request_body, &stream_fn](httplib::DataSink& route_sink) {
                    stream_fn(request_body, route_sink);
                });
            return false;
        });
}

int get_http_status_from_error(const std::string& error_code) {
    if (error_code == "slots_pinned_error" ||
        error_code == "router_residency_conflict") {
        return 409;
    } else if (error_code == "model_load_error") {
        return 500;
    } else {
        return 404;
    }
}

bool is_quiet_polling_path(const std::string& path) {
    return path == "/api/v0/downloads" || path == "/api/v1/downloads" ||
           path == "/v0/downloads" || path == "/v1/downloads" ||
           path == "/api/v0/system-stats" || path == "/api/v1/system-stats" ||
           path == "/v0/system-stats" || path == "/v1/system-stats" ||
           path == "/api/v0/stats" || path == "/api/v1/stats" ||
           path == "/v0/stats" || path == "/v1/stats" ||
           path == "/internal/models/sync" || path == "/internal/models/sync/status";
}

std::string join_warnings(const std::vector<std::string>& warnings) {
    std::ostringstream joined;
    for (size_t i = 0; i < warnings.size(); ++i) {
        if (i > 0) {
            joined << " | ";
        }
        joined << warnings[i];
    }
    return joined.str();
}

void attach_warnings(json& response, const std::vector<std::string>& warnings) {
    if (warnings.empty()) {
        return;
    }
    response["warnings"] = warnings;
    // Backward-compatible single-string field for older clients.
    response["warning"] = join_warnings(warnings);
}

nlohmann::json get_model_storage_stats(const std::string& model_storage_path) {
    auto make_error_result = [](const fs::path& path, const std::string& error) {
        return nlohmann::json{
            {"path", utils::path_to_utf8(path)},
            {"used_bytes", nullptr},
            {"total_bytes", nullptr},
            {"free_bytes", nullptr},
            {"error", error}
        };
    };

    std::error_code ec;
    fs::path configured_path;

    if (!model_storage_path.empty()) {
        configured_path = utils::path_from_utf8(model_storage_path);
    }

    if (configured_path.empty()) {
        configured_path = fs::current_path(ec);
        if (ec) {
            LOG(WARNING, "Server") << "Unable to resolve current path for model storage stats: "
                                   << ec.message() << std::endl;
            return make_error_result(
                fs::path{},
                "Unable to resolve current path: " + ec.message()
            );
        }
    } else if (configured_path.is_relative()) {
        configured_path = fs::absolute(configured_path, ec);
        if (ec) {
            LOG(WARNING, "Server") << "Unable to resolve model storage path "
                                   << model_storage_path << ": " << ec.message() << std::endl;
            return make_error_result(
                configured_path,
                "Unable to resolve model storage path: " + ec.message()
            );
        }
    }

    configured_path = configured_path.lexically_normal();

    fs::path probe_path = configured_path;
    while (!probe_path.empty()) {
        std::error_code exists_ec;
        if (fs::exists(probe_path, exists_ec)) {
            break;
        }

        if (exists_ec) {
            LOG(WARNING, "Server") << "Unable to inspect model storage path "
                                   << utils::path_to_utf8(probe_path) << ": "
                                   << exists_ec.message() << std::endl;
            return make_error_result(
                configured_path,
                "Unable to inspect model storage path: " + exists_ec.message()
            );
        }

        fs::path parent_path = probe_path.parent_path();
        if (parent_path == probe_path) {
            break;
        }

        probe_path = parent_path;
    }

    auto space_info = fs::space(probe_path, ec);
    if (ec) {
        LOG(WARNING, "Server") << "Unable to read model storage stats for "
                               << utils::path_to_utf8(probe_path) << ": "
                               << ec.message() << std::endl;
        return make_error_result(
            configured_path,
            "Unable to read model storage stats: " + ec.message()
        );
    }

    const uintmax_t total_bytes = space_info.capacity;
    const uintmax_t free_bytes = std::min(space_info.available, space_info.capacity);
    const uintmax_t used_bytes = total_bytes - free_bytes;

    return nlohmann::json{
        {"path", utils::path_to_utf8(configured_path)},
        {"used_bytes", static_cast<uint64_t>(used_bytes)},
        {"total_bytes", static_cast<uint64_t>(total_bytes)},
        {"free_bytes", static_cast<uint64_t>(free_bytes)}
    };
}

} // namespace


static const json MIME_TYPES = {
    {"mp3",  "audio/mpeg"},
    {"opus", "audio/opus"},
    {"aac",  "audio/aac"},
    {"flac", "audio/flac"},
    {"wav",  "audio/wav"},
    {"pcm",  "audio/l16;rate=24000;endianness=little-endian"}
};

Server::Server(std::shared_ptr<RuntimeConfig> config,
               const std::string& cache_dir,
               const std::string& config_dir)
    : config_(config),
      cache_dir_(cache_dir),
      config_dir_(config_dir),
      port_(config->port()), running_(false), udp_beacon_(),
      metrics_platform_(create_metrics_platform()) {

    // Set global HttpClient timeout
    utils::HttpClient::set_default_timeout(config->global_timeout());

    // Global download rate limit
    utils::HttpClient::set_download_rate_limit(config->download_rate_limit_bytes_per_second());

    cloud_registry_ = std::make_unique<CloudProviderRegistry>();
    // Seed installed providers from config.json. Runtime keys stay empty
    // until either an env var resolves them per-request or a client POSTs
    // /v1/cloud/auth — by design we never persist secrets to disk.
    {
        json snap = config_->snapshot();
        if (snap.contains("cloud_providers")) {
            cloud_registry_->load_from_config(snap["cloud_providers"]);
        }
    }

    alias_manager_ = std::make_unique<AliasManager>(cache_dir_);
    model_manager_ = std::make_unique<ModelManager>(config_->extra_models_dir());
    model_manager_->set_cloud_registry(cloud_registry_.get());

    backend_manager_ = std::make_unique<BackendManager>();
    BackendManager::set_global(backend_manager_.get());

    router_ = std::make_unique<Router>(config_.get(),
                                       model_manager_.get(),
                                       backend_manager_.get());
    router_->set_cloud_registry(cloud_registry_.get());

    // When a router collection is added, edited, or removed (via the API or an
    // on-disk edit), reclaim any routing helper no remaining policy references.
    model_manager_->set_models_changed_callback([this](uint64_t generation) {
        router_->reconcile_routing_helpers(active_policy_helper_models(), generation);
    });

    // Seed the router's needed-helper set from policies already present at
    // startup so a helper loaded before the first policy change still validates
    // against an authoritative set (see Router::load_model). Reserve the
    // generation BEFORE snapshotting the policies: the directory watcher may
    // already be publishing newer snapshots, and evaluating next_notify_generation()
    // after computing the snapshot (argument evaluation order is unspecified in
    // C++) could stamp this stale snapshot with a newer generation and clobber
    // the watcher's authoritative state.
    const uint64_t seed_generation = model_manager_->next_notify_generation();
    const std::set<std::string> seed_needed = active_policy_helper_models();
    router_->reconcile_routing_helpers(seed_needed, seed_generation);

    model_manager_->set_model_updated_callback([this](const std::string& model_name) {
        if (router_ && router_->is_model_loaded(model_name)) {
            LOG(INFO, "Server") << "Evicting updated model from memory: " << model_name << std::endl;
            router_->unload_model(model_name);
        }
    });
    {
        lemon::jobs::OpProviders providers;
        struct JobModelState {
            std::map<std::string, bool> snapshot;
            std::map<std::string, nlohmann::json> owned;
        };
        auto job_states = std::make_shared<std::map<std::string, JobModelState>>();
        auto current_job = std::make_shared<std::string>();
        auto state_mutex = std::make_shared<std::mutex>();
        providers.system_info = [] {
            return lemon::jobs::json::parse(SystemInfoCache::get_system_info_with_cache().dump());
        };
        providers.system_stats = [this] {
            lemon::jobs::json stats;
            const double cpu_percent = get_cpu_usage();
            stats["cpu_percent"] =
                cpu_percent >= 0 ? lemon::jobs::json(cpu_percent) : lemon::jobs::json();
            stats["memory_gb"] = metrics_platform_->get_memory_usage_gb();
            const double gpu_percent = get_gpu_usage();
            stats["gpu_percent"] =
                gpu_percent >= 0 ? lemon::jobs::json(gpu_percent) : lemon::jobs::json();
            const double vram_gb = get_vram_usage();
            stats["vram_gb"] = vram_gb >= 0 ? lemon::jobs::json(vram_gb) : lemon::jobs::json();
            const double npu_percent = get_npu_utilization();
            stats["npu_percent"] =
                npu_percent >= 0 ? lemon::jobs::json(npu_percent) : lemon::jobs::json();
            return stats;
        };
        providers.models_list = [this] {
            nlohmann::json data = nlohmann::json::array();
            for (const auto& [model_id, info] : model_manager_->get_supported_models())
                data.push_back(model_info_to_json(model_id, info));
            nlohmann::json response = {{"object", "list"}, {"data", data}};
            return lemon::jobs::json::parse(response.dump());
        };
        providers.model_get = [this](const std::string& id) -> lemon::jobs::json {
            auto models = model_manager_->get_supported_models();
            auto it = models.find(id);
            if (it == models.end()) return lemon::jobs::json(nullptr);
            return lemon::jobs::json::parse(model_info_to_json(id, it->second).dump());
        };
        providers.load_op = [this, job_states, current_job, state_mutex](
                                const lemon::jobs::json& params,
                                lemon::jobs::CancelFlag& cancel) -> lemon::jobs::json {
            if (!params.contains("model") || !params["model"].is_string())
                throw lemon::jobs::JobError(400, "load requires a 'model' string");
            const std::string model = params["model"].get<std::string>();
            if (!model_manager_->model_exists(model))
                throw lemon::jobs::JobError(404, "unknown model '" + model + "'");
            if (!model_manager_->is_model_downloaded(model))
                throw lemon::jobs::JobError(404, "model '" + model + "' is not downloaded");
            auto info = model_manager_->get_model_info(model);
            nlohmann::json opt_json = nlohmann::json::parse(params.dump());
            RecipeOptions options(info.recipe, opt_json);
            std::optional<bool> pinned = std::nullopt;
            if (params.contains("pinned") && params["pinned"].is_boolean())
                pinned = params["pinned"].get<bool>();
            try {
                router_->load_model(model, info, options, true, true, pinned,
                                    LoadPurpose::UserInference, &cancel);
            } catch (const std::exception& e) {
                throw lemon::jobs::JobError(500, e.what());
            }
            {
                const std::string canonical = router_->canonical_model_name(model);
                const int pid = router_->loaded_model_pid(model);
                std::lock_guard<std::mutex> lk(*state_mutex);
                if (!current_job->empty()) {
                    auto it = job_states->find(*current_job);
                    if (it != job_states->end())
                        it->second.owned[canonical] = {{"pid", pid}};
                }
            }
            if (cancel.load()) {
                throw lemon::jobs::JobError(499, "interrupted");
            }
            RecipeOptions effective = router_->get_model_recipe_options(model);
            lemon::jobs::json out;
            out["loaded"] = true;
            out["model"] = model;
            const nlohmann::json backend_json = effective.get_option(info.recipe + "_backend");
            if (backend_json.is_string()) out["backend"] = backend_json.get<std::string>();
            const nlohmann::json ctx_json = effective.get_option("ctx_size");
            if (ctx_json.is_number()) out["ctx_size"] = ctx_json.get<int64_t>();
            return out;
        };
        providers.unload_op = [this, job_states, current_job, state_mutex](
                                  const lemon::jobs::json& params,
                                  lemon::jobs::CancelFlag&) -> lemon::jobs::json {
            std::string model;
            if (params.contains("model") && params["model"].is_string())
                model = params["model"].get<std::string>();
            if (!model.empty()) {
                if (router_->is_model_loaded(model)) router_->unload_model(model);
            } else {
                router_->unload_model("");
            }
            {
                std::lock_guard<std::mutex> lk(*state_mutex);
                if (!current_job->empty()) {
                    auto it = job_states->find(*current_job);
                    if (it != job_states->end()) {
                        if (model.empty()) it->second.owned.clear();
                        else it->second.owned.erase(router_->canonical_model_name(model));
                    }
                }
            }
            return lemon::jobs::json::object();
        };
        providers.chat_op = [this](const lemon::jobs::json& params,
                                   lemon::jobs::CancelFlag& cancel) -> lemon::jobs::json {
            nlohmann::json request = nlohmann::json::parse(params.dump());
            nlohmann::json response = router_->chat_completion(request, &cancel);
            if (response.contains("error")) {
                std::string msg = "chat failed";
                const auto& err = response["error"];
                if (err.is_object() && err.contains("message") && err["message"].is_string())
                    msg = err["message"].get<std::string>();
                else if (err.is_string())
                    msg = err.get<std::string>();
                throw lemon::jobs::JobError(424, msg);
            }
            return lemon::jobs::json::parse(response.dump());
        };
        providers.begin_exclusive = [this, job_states, current_job, state_mutex](
                                        const std::string& job_id,
                                        lemon::jobs::CancelFlag* cancel) -> bool {
            if (!router_->begin_exclusive(cancel)) return false;
            std::lock_guard<std::mutex> lk(*state_mutex);
            if (!job_states->count(job_id))
                (*job_states)[job_id].snapshot = router_->snapshot_loaded_models();
            *current_job = job_id;
            return true;
        };
        providers.end_exclusive = [this, current_job, state_mutex] {
            {
                std::lock_guard<std::mutex> lk(*state_mutex);
                current_job->clear();
            }
            router_->end_exclusive();
        };
        providers.reconcile_unload = [this, job_states, state_mutex](
                                         const std::string& job_id) {
            std::map<std::string, int> owned_live;
            std::map<std::string, bool> snapshot;
            {
                std::lock_guard<std::mutex> lk(*state_mutex);
                auto it = job_states->find(job_id);
                if (it == job_states->end()) return;
                for (const auto& kv : it->second.owned)
                    if (kv.second.is_object() && kv.second.contains("pid"))
                        owned_live[kv.first] = kv.second["pid"].get<int>();
                snapshot = it->second.snapshot;
            }
            if (owned_live.empty()) return;
            auto captured = router_->unload_job_models(owned_live, snapshot);
            std::lock_guard<std::mutex> lk(*state_mutex);
            auto it = job_states->find(job_id);
            if (it == job_states->end()) return;
            for (auto& kv : captured) it->second.owned[kv.first] = std::move(kv.second);
        };
        providers.restore_exclusive = [this, job_states, state_mutex](
                                          const std::string& job_id,
                                          const lemon::jobs::json& manifest,
                                          lemon::jobs::CancelFlag* cancel) -> bool {
            std::map<std::string, nlohmann::json> to_restore;
            std::set<std::string> tracked;
            {
                std::lock_guard<std::mutex> lk(*state_mutex);
                auto it = job_states->find(job_id);
                if (it != job_states->end()) {
                    for (const auto& kv : it->second.owned) {
                        tracked.insert(kv.first);
                        if (kv.second.is_object() && kv.second.contains("options"))
                            to_restore[kv.first] = kv.second;
                    }
                }
            }
            for (auto it = manifest.begin(); it != manifest.end(); ++it) {
                const std::string canonical = router_->canonical_model_name(it.key());
                if (tracked.count(canonical) || to_restore.count(canonical)) continue;
                nlohmann::json params = nlohmann::json::parse(it.value().dump());
                nlohmann::json entry = {{"options", params}};
                if (params.contains("pinned") && params["pinned"].is_boolean())
                    entry["pinned"] = params["pinned"].get<bool>();
                to_restore[canonical] = std::move(entry);
            }
            for (const auto& kv : to_restore) {
                if (cancel && cancel->load()) return false;
                if (router_->is_model_loaded(kv.first)) {
                    std::lock_guard<std::mutex> lk(*state_mutex);
                    auto it = job_states->find(job_id);
                    if (it != job_states->end()) it->second.owned.erase(kv.first);
                    continue;
                }
                try {
                    auto info = model_manager_->get_model_info(kv.first);
                    RecipeOptions options(info.recipe, kv.second.value("options", nlohmann::json::object()));
                    std::optional<bool> pinned = std::nullopt;
                    if (kv.second.contains("pinned") && kv.second["pinned"].is_boolean())
                        pinned = kv.second["pinned"].get<bool>();
                    router_->load_model(kv.first, info, options, true, true, pinned,
                                        LoadPurpose::UserInference, cancel);
                    const int pid = router_->loaded_model_pid(kv.first);
                    std::lock_guard<std::mutex> lk(*state_mutex);
                    auto it = job_states->find(job_id);
                    if (it != job_states->end()) it->second.owned[kv.first] = {{"pid", pid}};
                } catch (const std::exception& e) {
                    if (cancel && cancel->load()) return false;
                    LOG(WARNING, "Jobs") << "could not restore job model '" << kv.first
                                         << "' on resume: " << e.what() << std::endl;
                    std::lock_guard<std::mutex> lk(*state_mutex);
                    auto it = job_states->find(job_id);
                    if (it != job_states->end()) it->second.owned.erase(kv.first);
                }
            }
            return true;
        };
        providers.discard_exclusive = [job_states, state_mutex](
                                          const std::string& job_id) {
            std::lock_guard<std::mutex> lk(*state_mutex);
            job_states->erase(job_id);
        };
        job_manager_ = std::make_unique<lemon::jobs::JobManager>(
            cache_dir_, config_dir_, lemon::jobs::build_op_registry(std::move(providers)));
    }

    LOG(DEBUG, "Server") << "Debug logging enabled - subprocess output will be visible" << std::endl;

    const char* api_key_env = std::getenv("LEMONADE_API_KEY");
    api_key_ = api_key_env ? std::string(api_key_env) : "";

    // Read admin API key - if not set, defaults to regular API key value
    const char* admin_api_key_env = std::getenv("LEMONADE_ADMIN_API_KEY");
    if (admin_api_key_env) {
        admin_api_key_ = std::string(admin_api_key_env);
    } else {
        admin_api_key_ = api_key_;
    }

    setup_http_servers();

    // Initialize WebSocket server for realtime API and log streaming
    websocket_server_ = std::make_unique<WebSocketServer>(
        router_.get(),
        config_->host(),
        config_->websocket_port());

    start_model_cache_warmup();
}

void Server::start_model_cache_warmup() {
    if (model_cache_warmup_thread_.joinable()) {
        return;
    }

    model_cache_warmup_thread_ = std::thread([this]() {
        try {
            LOG(DEBUG, "Server") << "Warming model list cache..." << std::endl;
            model_manager_->get_supported_models();
            LOG(DEBUG, "Server") << "Model list cache warmup complete" << std::endl;
        } catch (const std::exception& e) {
            LOG(WARNING, "Server") << "Model list cache warmup failed: " << e.what() << std::endl;
        } catch (...) {
            LOG(WARNING, "Server") << "Model list cache warmup failed with unknown error" << std::endl;
        }

        if (config_->auto_check_model_updates()) {
            try {
                LOG(DEBUG, "Server") << "Checking downloaded models for updates..." << std::endl;
                auto check_res = model_manager_->check_for_model_updates();
                LOG(DEBUG, "Server") << "Model update check complete" << std::endl;

                for (const auto& [model_name, err] : check_res.failed_models) {
                    LOG(WARNING, "Server") << "Model update check failed for " << model_name << ": " << err << std::endl;
                }

                std::vector<std::string> auto_update_targets;
                for (const auto& public_name : check_res.updated_models) {

                    try {
                        std::string canonical_name = model_manager_->resolve_model_name(public_name);
                        ModelInfo info = model_manager_->get_model_info(canonical_name);
                        if (model_manager_->should_auto_update(info)) {
                            auto_update_targets.push_back(public_name);
                        }
                    } catch (...) {}
                }


                if (!auto_update_targets.empty()) {
                    LOG(INFO, "Server") << "Auto-updating " << auto_update_targets.size() << " model(s)..." << std::endl;
                    (void)model_manager_->sync_models(auto_update_targets, /*dry_run=*/false);
                }
            } catch (const std::exception& e) {
                LOG(WARNING, "Server") << "Model update check failed: " << e.what() << std::endl;
            } catch (...) {
                LOG(WARNING, "Server") << "Model update check failed with unknown error" << std::endl;
            }
        } else {
            LOG(DEBUG, "Server")
                << "Automatic model update checks are disabled" << std::endl;
        }

        update_check_done_ = true;
    });
}

// Extract the member-function pointer for httplib::Server's private virtual
// process_and_close_socket (see upgradable_http_server.h). Explicit
// instantiation is the one context where C++ permits naming a private member.
template struct lemon::detail::PrivateMemberInit<
    lemon::detail::ProcessAndCloseSocketTag,
    &httplib::Server::process_and_close_socket>;

void Server::setup_http_servers() {
    http_server_ = std::make_unique<RoutedHttpServer>();
    http_server_v6_ = std::make_unique<RoutedHttpServer>();

    // Front listeners for the main port: WebSocket upgrades for /realtime and
    // /logs/stream are adopted by the libwebsockets server; everything else is
    // processed by the routed servers above. The dedicated websocket_port
    // listener keeps running unchanged.
    auto upgrade_handler = [this](socket_t sock) -> bool {
        if (websocket_server_ && websocket_server_->is_running()) {
            return websocket_server_->adopt_socket(static_cast<intptr_t>(sock));
        }
        return false;
    };
    http_front_ = std::make_unique<UpgradableFrontServer>(http_server_.get(), upgrade_handler);
    http_front_v6_ = std::make_unique<UpgradableFrontServer>(http_server_v6_.get(), upgrade_handler);

    // Keep cpp-httplib's default socket options here. httplib binds IPv6 with
    // IPV6_V6ONLY=0, so "::" overlaps the IPv4 wildcard "0.0.0.0" and only the
    // default SO_REUSEPORT lets the two coexist. Duplicate detection is done by
    // port_is_available() in run(), not by making these listeners exclusive.

    // CRITICAL: Enable multi-threading so the server can handle concurrent requests
    // Without this, the server is single-threaded and blocks on long operations

    // Size the pool from the host CPU count instead of a fixed 8. cpp-httplib
    // dedicates one worker thread per in-flight request for the connection's
    // lifetime, so a small fixed pool lets a handful of slow-loris or long-lived
    // streaming connections starve the management endpoints (/health, /load).
    unsigned int hw = std::thread::hardware_concurrency();
    size_t thread_count = std::clamp<size_t>(static_cast<size_t>(hw) * 4, 32, 256);
    std::function<httplib::TaskQueue *(void)> task_queue_factory = [thread_count] {
        LOG(DEBUG, "Server") << "Creating new thread pool with " << thread_count
                             << " threads" << std::endl;
        return new httplib::ThreadPool(thread_count);
    };

    // The fronts own the accept loops (and therefore the task queues)
    http_front_->new_task_queue = task_queue_factory;
    http_front_v6_->new_task_queue = task_queue_factory;
    http_server_->new_task_queue = task_queue_factory;
    http_server_v6_->new_task_queue = task_queue_factory;

    // Bound how long a single connection can tie up a worker thread. Without a
    // read timeout a client that opens a socket and never finishes its request
    // (slow loris) holds its worker indefinitely. Streaming responses drive
    // their own write cadence, so keep the write timeout generous. The fronts
    // need the same limits: they own accept and run the WebSocket upgrade peek
    // before delegating, so a stalled client could otherwise hold a front
    // worker there.
    for (httplib::Server* srv : {static_cast<httplib::Server*>(http_front_.get()),
                                 static_cast<httplib::Server*>(http_front_v6_.get()),
                                 static_cast<httplib::Server*>(http_server_.get()),
                                 static_cast<httplib::Server*>(http_server_v6_.get())}) {
        srv->set_read_timeout(30, 0);
        srv->set_write_timeout(300, 0);
        srv->set_keep_alive_max_count(100);
    }

    setup_routes(*http_server_);
    setup_routes(*http_server_v6_);
}

void Server::stop_http_listeners() {
    // The routed servers never own the listen socket: clear the injected fd so
    // their per-connection keep-alive loops exit, then close it once via the
    // fronts (which are the servers actually listening).
    if (http_server_) {
        http_server_->set_listen_socket(INVALID_SOCKET);
    }
    if (http_server_v6_) {
        http_server_v6_->set_listen_socket(INVALID_SOCKET);
    }
    if (http_front_) {
        http_front_->stop();
    }
    if (http_front_v6_) {
        http_front_v6_->stop();
    }
}

Server::~Server() {
    cancel_download_jobs();
    if (model_manager_) {
        model_manager_->cancel_sync();
    }
    stop();
    if (model_cache_warmup_thread_.joinable()) {
        model_cache_warmup_thread_.join();
    }

    std::lock_guard<std::mutex> lock(background_sync_mutex_);
    for (auto& entry : background_sync_threads_) {
        if (entry.thread.joinable()) {
            entry.thread.join();
        }
    }
    background_sync_threads_.clear();
}



namespace {

// Parse a W3C Trace Context "traceparent" header per the spec grammar
// (https://www.w3.org/TR/trace-context/): version "-" trace-id "-" parent-id
// "-" trace-flags, where version is 2 lowercase hex chars, trace-id is 32,
// parent-id is 16, and trace-flags is 2. The all-zero trace-id and parent-id
// are invalid. On success, out_trace_id/out_parent_id receive the lowercase
// hex ids and the function returns true; otherwise returns false.
bool parse_traceparent(const std::string& header, std::string& out_trace_id, std::string& out_parent_id) {
    // A version-00 traceparent is exactly 55 chars. Reject anything shorter and
    // cap the length so a header packed with dashes can't drive an unbounded
    // split loop; the upper bound leaves room for higher versions that may
    // append trailing fields (handled below).
    if (header.size() < 55 || header.size() > 128) return false;

    auto is_hex = [](const std::string& s) {
        return !s.empty() && std::all_of(s.begin(), s.end(), [](unsigned char c) {
            return (c >= '0' && c <= '9') || (c >= 'a' && c <= 'f');
        });
    };
    auto is_all_zero = [](const std::string& s) {
        return s.find_first_not_of('0') == std::string::npos;
    };

    // Lowercase a copy so uppercase hex from lenient clients still validates.
    std::string h = header;
    std::transform(h.begin(), h.end(), h.begin(), [](unsigned char c) { return std::tolower(c); });

    std::vector<std::string> parts;
    size_t start = 0;
    while (true) {
        size_t dash = h.find('-', start);
        if (dash == std::string::npos) {
            parts.push_back(h.substr(start));
            break;
        }
        parts.push_back(h.substr(start, dash - start));
        start = dash + 1;
    }

    if (parts.size() < 4) return false;
    const std::string& version = parts[0];
    const std::string& trace_id = parts[1];
    const std::string& parent_id = parts[2];
    const std::string& flags = parts[3];

    if (version.size() != 2 || !is_hex(version) || version == "ff") return false;
    // Version 00 carries exactly four fields; per W3C forward-compatibility a
    // higher version may append trailing fields, which we parse then ignore.
    if (version == "00" && parts.size() != 4) return false;
    if (trace_id.size() != 32 || !is_hex(trace_id) || is_all_zero(trace_id)) return false;
    if (parent_id.size() != 16 || !is_hex(parent_id) || is_all_zero(parent_id)) return false;
    if (flags.size() != 2 || !is_hex(flags)) return false;

    out_trace_id = trace_id;
    out_parent_id = parent_id;
    return true;
}

} // namespace

void Server::log_request(const httplib::Request& req) {
    if (req.path != "/api/v0/health" && req.path != "/api/v1/health" &&
        req.path != "/v0/health" && req.path != "/v1/health" &&
        req.path != "/live" &&
        req.path != "/metrics" &&
        !is_quiet_polling_path(req.path)) {
        LOG(DEBUG, "Server") << req.method << " " << req.path << std::endl;
    }
}

httplib::Server::HandlerResponse Server::authenticate_request(const httplib::Request& req, httplib::Response& res) {
    telemetry::g_request_start_time = std::chrono::steady_clock::now();
    telemetry::g_current_auth_token = "";
    if (req.has_header("X-Client-Session-Id")) {
        telemetry::g_current_client_session_id = req.get_header_value("X-Client-Session-Id");
    } else {
        telemetry::g_current_client_session_id.clear();
    }

    // Opt-in W3C Trace Context ingestion: when enabled, adopt a valid incoming
    // "traceparent" so inference spans join the caller's distributed trace
    // instead of starting a fresh root. Cleared otherwise so a stale thread-local
    // never leaks across requests on a reused worker thread.
    telemetry::g_incoming_trace_id.clear();
    telemetry::g_incoming_parent_span_id.clear();
    if (config_->telemetry_trust_incoming_trace_context() && req.has_header("traceparent")) {
        std::string trace_id;
        std::string parent_id;
        if (parse_traceparent(req.get_header_value("traceparent"), trace_id, parent_id)) {
            telemetry::g_incoming_trace_id = trace_id;
            telemetry::g_incoming_parent_span_id = parent_id;
        }
    }

    telemetry::g_incoming_client_id.clear();
    telemetry::g_incoming_session_id.clear();

    auto trim = [](const std::string& str) -> std::string {
        size_t s = str.find_first_not_of(" \t\r\n");
        if (s == std::string::npos) return "";
        size_t e = str.find_last_not_of(" \t\r\n");
        return str.substr(s, e - s + 1);
    };

    std::string client_val;
    if (config_) {
        for (const auto& hdr : config_->telemetry_session_headers_client()) {
            std::string cleaned_hdr = trim(hdr);
            if (!cleaned_hdr.empty() && req.has_header(cleaned_hdr)) {
                std::string val = trim(req.get_header_value(cleaned_hdr));
                if (!val.empty()) {
                    client_val = val;
                    break;
                }
            }
        }
    }
    if (client_val.empty()) {
        static const char* kWellKnownClientHeaders[] = {
            "x-opencode-client",
            "x-client-id",
            "x-client-name"
        };
        for (const char* hdr : kWellKnownClientHeaders) {
            if (req.has_header(hdr)) {
                std::string val = trim(req.get_header_value(hdr));
                if (!val.empty()) {
                    client_val = val;
                    break;
                }
            }
        }
    }

    std::string session_val;
    if (config_) {
        for (const auto& hdr : config_->telemetry_session_headers_id()) {
            std::string cleaned_hdr = trim(hdr);
            if (!cleaned_hdr.empty() && req.has_header(cleaned_hdr)) {
                std::string val = trim(req.get_header_value(cleaned_hdr));
                if (!val.empty()) {
                    session_val = val;
                    break;
                }
            }
        }
    }
    if (session_val.empty()) {
        static const char* kWellKnownSessionHeaders[] = {
            "x-opencode-session",
            "x-session-id",
            "x-client-session-id",
            "mcp-session-id",
            "x-conversation-id",
            "session-id"
        };
        for (const char* hdr : kWellKnownSessionHeaders) {
            if (req.has_header(hdr)) {
                std::string val = trim(req.get_header_value(hdr));
                if (!val.empty()) {
                    session_val = val;
                    break;
                }
            }
        }
    }

    if (!session_val.empty()) {
        telemetry::g_incoming_client_id = client_val;
        telemetry::g_incoming_session_id = session_val;
    }

    // Check if path requires authentication (API routes and internal endpoints).
    // /mcp is included here so that LEMONADE_API_KEY enforcement covers the MCP
    // gateway (Critical Invariant #10). It is the only API route outside the
    // /api/, /v0/, /v1/ prefixes — see register_routes() in McpServer for why.
    bool is_api_route = (req.path.rfind("/api/", 0) == 0) ||
                        (req.path.rfind("/v0/", 0) == 0) ||
                        (req.path.rfind("/v1/", 0) == 0) ||
                        (req.path == "/mcp");
    bool is_internal_route = (req.path.rfind("/internal/", 0) == 0);
    bool is_metrics_route = (req.path == "/metrics");

    // Authentication hierarchy. Two credentials gate two classes of endpoints:
    // api_key_ gates the regular API endpoints (/api, /v0, /v1); admin_api_key_
    // gates the internal control endpoints (/internal/*). admin_api_key_ defaults
    // to api_key_ when LEMONADE_ADMIN_API_KEY is unset.
    // - admin_api_key_ authenticates against both regular and internal endpoints.
    // - api_key_ authenticates against the regular endpoints only. It cannot
    //   reach /internal/* when LEMONADE_ADMIN_API_KEY is set to a distinct value;
    //   when LEMONADE_ADMIN_API_KEY is unset, admin_api_key_ == api_key_, so the
    //   regular key also authenticates against /internal/*.
    // - If api_key_ is empty, the regular endpoints require no authentication.
    // - If admin_api_key_ is empty (neither key set), legacy /internal/* routes
    //   require no authentication. The MCP process-launch surface is deliberately
    //   fail-closed and requires an explicitly configured admin key.

    // Safely extract bearer token, guarding against malformed Authorization headers
    std::string auth_token;
    try {
        if (req.has_header("Authorization")) {
            auto auth_value = req.get_header_value("Authorization");
            // httplib::get_bearer_token_auth does substr(7) for "Bearer ", so check length
            if (auth_value.size() >= 7) {
                auth_token = httplib::get_bearer_token_auth(req);
            }
            // Silently ignore malformed/short Authorization headers
        }
    } catch (const std::exception& e) {
        LOG(DEBUG, "Server") << "Failed to parse Authorization header: " << e.what() << std::endl;
    }

    telemetry::g_current_auth_token = auth_token;

    const bool is_mcp_internal_route =
        req.path == "/internal/mcp" ||
        req.path.rfind("/internal/mcp/", 0) == 0;

    if (is_internal_route) {
        // MCP server registration can launch arbitrary local processes. Do not
        // expose that capability on a keyless server, even on loopback: permissive
        // CORS would otherwise let an unrelated web page drive these endpoints.
        // Apply this to OPTIONS as well so a browser preflight fails closed.
        if (is_mcp_internal_route && admin_api_key_.empty()) {
            res.status = 403;
            res.set_content(
                "{\"error\": \"MCP administration requires LEMONADE_ADMIN_API_KEY or LEMONADE_API_KEY\"}",
                "application/json");
            return httplib::Server::HandlerResponse::Handled;
        }

        // Internal routes require admin key authentication
        if (!admin_api_key_.empty() && req.method != "OPTIONS") {
            if (auth_token != admin_api_key_) {
                res.status = 401;
                res.set_content("{\"error\": \"Invalid or missing admin API key\"}", "application/json");
                return httplib::Server::HandlerResponse::Handled;
            }
        }
    } else if ((is_api_route || is_metrics_route) && req.method != "OPTIONS") {
        if (!api_key_.empty()) {
            if ((auth_token != api_key_) && (auth_token != admin_api_key_)) {
                res.status = 401;
                res.set_content("{\"error\": \"Invalid or missing API key\"}", "application/json");
                return httplib::Server::HandlerResponse::Handled;
            }
        }
    }

    return httplib::Server::HandlerResponse::Unhandled;
}


void Server::setup_routes(httplib::Server &web_server) {
    // Add pre-routing handler to log ALL incoming requests (except health checks)
    web_server.set_pre_routing_handler([this](const httplib::Request& req, httplib::Response& res) {
        this->log_request(req);

        // Unconditionally set Vary: Origin to prevent caching issues, preserving existing values
        std::string vary = "Origin";
        if (res.has_header("Vary")) {
            std::string existing = res.get_header_value("Vary");
            if (existing.find("Origin") == std::string::npos) {
                vary = existing + ", Origin";
            } else {
                vary = existing;
            }
        }
        res.set_header("Vary", vary);

        if (req.has_header("Origin")) {
            std::string origin = req.get_header_value("Origin");
            const char* env_origins = std::getenv("LEMONADE_ALLOWED_ORIGINS");
            std::string allowed_origins = env_origins ? std::string(env_origins) : "";

            if (utils::is_origin_allowed(origin, allowed_origins)) {
                res.set_header("Access-Control-Allow-Origin", origin);
                if (req.has_header("Access-Control-Request-Private-Network") &&
                    req.get_header_value("Access-Control-Request-Private-Network") == "true") {
                    res.set_header("Access-Control-Allow-Private-Network", "true");
                }
            } else {
                res.status = 403;
                res.set_content("{\"error\": \"Origin not allowed\"}", "application/json");
                return httplib::Server::HandlerResponse::Handled;
            }
        }

        return authenticate_request(req, res);
    });

    web_server.Get("/live", [this](const httplib::Request& req, httplib::Response& res) {
        handle_live(req, res);
    });

    // Prometheus scrape endpoint for Lemonade, model, backend, and system metrics.
    web_server.Get("/metrics", [this](const httplib::Request& req, httplib::Response& res) {
        handle_metrics(req, res);
    });

    // Setup CORS for all routes
    setup_cors(web_server);

    // Helper lambda to register routes for both v0 and v1 (with and without /api prefix for OpenAI compatibility)
    auto register_get = [this, &web_server](const std::string& endpoint,
                               std::function<void(const httplib::Request&, httplib::Response&)> handler) {
        web_server.Get("/api/v0/" + endpoint, handler);
        web_server.Get("/api/v1/" + endpoint, handler);
        web_server.Get("/v0/" + endpoint, handler);
        web_server.Get("/v1/" + endpoint, handler);
    };

    auto register_post = [this, &web_server](const std::string& endpoint,
                                std::function<void(const httplib::Request&, httplib::Response&)> handler) {
        web_server.Post("/api/v0/" + endpoint, handler);
        web_server.Post("/api/v1/" + endpoint, handler);
        web_server.Post("/v0/" + endpoint, handler);
        web_server.Post("/v1/" + endpoint, handler);
        if (endpoint != "jobs") {
            web_server.Get("/api/v0/" + endpoint, [](const httplib::Request&, httplib::Response& res) {
                res.status = 405;
                res.set_content("{\"error\": \"Method Not Allowed. Use POST for this endpoint\"}", "application/json");
            });
            web_server.Get("/api/v1/" + endpoint, [](const httplib::Request&, httplib::Response& res) {
                res.status = 405;
                res.set_content("{\"error\": \"Method Not Allowed. Use POST for this endpoint\"}", "application/json");
            });
            web_server.Get("/v0/" + endpoint, [](const httplib::Request&, httplib::Response& res) {
                res.status = 405;
                res.set_content("{\"error\": \"Method Not Allowed. Use POST for this endpoint\"}", "application/json");
            });
            web_server.Get("/v1/" + endpoint, [](const httplib::Request&, httplib::Response& res) {
                res.status = 405;
                res.set_content("{\"error\": \"Method Not Allowed. Use POST for this endpoint\"}", "application/json");
            });
        }
    };

    // Health check
    register_get("health", [this](const httplib::Request& req, httplib::Response& res) {
        handle_health(req, res);
    });

    // Models endpoints
    register_get("models", [this](const httplib::Request& req, httplib::Response& res) {
        handle_models(req, res);
    });

    // Explicit network action for users who disable startup update checks.
    register_post("models/check-updates", [this](const httplib::Request& req, httplib::Response& res) {
        handle_model_update_check(req, res);
    });

    // Model files endpoint for the Files tab. Register before the generic
    // /models/(.+) route so '<model-id>/files' is not parsed as the model ID.
    web_server.Get(R"(/api/v0/models/(.+)/files)", [this](const httplib::Request& req, httplib::Response& res) {
        handle_model_files(req, res);
    });
    web_server.Get(R"(/api/v1/models/(.+)/files)", [this](const httplib::Request& req, httplib::Response& res) {
        handle_model_files(req, res);
    });
    web_server.Get(R"(/v0/models/(.+)/files)", [this](const httplib::Request& req, httplib::Response& res) {
        handle_model_files(req, res);
    });
    web_server.Get(R"(/v1/models/(.+)/files)", [this](const httplib::Request& req, httplib::Response& res) {
        handle_model_files(req, res);
    });

    // Per-model recipe options. Same ordering constraint as /files above: must
    // precede the generic /models/(.+) route.
    for (const char* prefix : {"/api/v0", "/api/v1", "/v0", "/v1"}) {
        const std::string route = std::string(prefix) + R"(/models/(.+)/options)";
        web_server.Get(route, [this](const httplib::Request& req, httplib::Response& res) {
            handle_model_options_get(req, res);
        });
        web_server.Post(route, [this](const httplib::Request& req, httplib::Response& res) {
            handle_model_options_post(req, res);
        });
        web_server.Delete(route, [this](const httplib::Request& req, httplib::Response& res) {
            handle_model_options_delete(req, res);
        });
    }

    // Model by ID (need to register for both versions with regex, with and without /api prefix)
    web_server.Get(R"(/api/v0/models/(.+))", [this](const httplib::Request& req, httplib::Response& res) {
        handle_model_by_id(req, res);
    });
    web_server.Get(R"(/api/v1/models/(.+))", [this](const httplib::Request& req, httplib::Response& res) {
        handle_model_by_id(req, res);
    });
    web_server.Get(R"(/v0/models/(.+))", [this](const httplib::Request& req, httplib::Response& res) {
        handle_model_by_id(req, res);
    });
    web_server.Get(R"(/v1/models/(.+))", [this](const httplib::Request& req, httplib::Response& res) {
        handle_model_by_id(req, res);
    });

    // Chat completions (OpenAI compatible)
    register_post("chat/completions", [this](const httplib::Request& req, httplib::Response& res) {
        handle_chat_completions(req, res);
    });

    // Completions
    register_post("completions", [this](const httplib::Request& req, httplib::Response& res) {
        handle_completions(req, res);
    });

    // Embeddings
    register_post("embeddings", [this](const httplib::Request& req, httplib::Response& res) {
        handle_embeddings(req, res);
    });

    // Reranking.
    // `/rerank` is the de-facto convention that most clients expect.
    // `/reranking` (Lemonade's original path) is kept as an alias for backward compatibility.
    // `/reranker` is added as an additional alias.
    for (const char* rerank_path : {"rerank", "reranking", "reranker"}) {
        register_post(rerank_path, [this](const httplib::Request& req, httplib::Response& res) {
            handle_reranking(req, res);
        });
    }

    register_post("classify", [this](const httplib::Request& req, httplib::Response& res) {
        handle_classify(req, res);
    });

    // Slots (llama.cpp backend information)
    register_get("slots", [this](const httplib::Request& req, httplib::Response& res) {
        handle_slots(req, res);
    });

    // Slots action endpoints (need to register for both versions with regex, with and without /api prefix)
    web_server.Post(R"(/api/v0/slots/(\d+))", [this](const httplib::Request& req, httplib::Response& res) {
        handle_slots_by_id(req, res);
    });
    web_server.Post(R"(/api/v1/slots/(\d+))", [this](const httplib::Request& req, httplib::Response& res) {
        handle_slots_by_id(req, res);
    });
    web_server.Post(R"(/v0/slots/(\d+))", [this](const httplib::Request& req, httplib::Response& res) {
        handle_slots_by_id(req, res);
    });
    web_server.Post(R"(/v1/slots/(\d+))", [this](const httplib::Request& req, httplib::Response& res) {
        handle_slots_by_id(req, res);
    });

    // Tokenize endpoint (llama.cpp specific)
    register_post("tokenize", [this](const httplib::Request& req, httplib::Response& res) {
        handle_tokenize(req, res);
    });

    // Audio endpoints (OpenAI /v1/audio/* compatible)
    register_post("audio/transcriptions", [this](const httplib::Request& req, httplib::Response& res) {
        handle_audio_transcriptions(req, res);
    });

    // Speech
    register_post("audio/speech", [this](const httplib::Request& req, httplib::Response& res) {
        handle_audio_speech(req, res);
    });

    // Image endpoints (OpenAI /v1/images/* compatible)
    register_post("images/generations", [this](const httplib::Request& req, httplib::Response& res) {
        handle_image_generations(req, res);
    });
    register_post("images/edits", [this](const httplib::Request& req, httplib::Response& res) {
        handle_image_edits(req, res);
    });
    register_post("images/variations", [this](const httplib::Request& req, httplib::Response& res) {
        handle_image_variations(req, res);
    });
    register_post("images/upscale", [this](const httplib::Request& req, httplib::Response& res) {
        handle_image_upscale(req, res);
    });
    // Generative-audio endpoint: text -> audio clip (music, sound effects)
    register_post("audio/generations", [this](const httplib::Request& req, httplib::Response& res) {
        handle_audio_generations(req, res);
    });
    register_post("3d/generations", [this](const httplib::Request& req, httplib::Response& res) {
        handle_3d_generations(req, res);
    });
    // Responses endpoint
    register_post("responses", [this](const httplib::Request& req, httplib::Response& res) {
        handle_responses(req, res);
    });

    // Model management endpoints
    register_post("models/register", [this](const httplib::Request& req, httplib::Response& res) {
        handle_model_register(req, res);
    });

    register_post("pull", [this](const httplib::Request& req, httplib::Response& res) {
        handle_pull(req, res);
    });

    register_get("registry/search", [this](const httplib::Request& req, httplib::Response& res) {
        handle_registry_search(req, res);
    });

    register_get("pull/variants", [this](const httplib::Request& req, httplib::Response& res) {
        handle_pull_variants(req, res);
    });

    // Runs a routing policy engine against an ad-hoc policy JSON + prompt
    // (the Router Builder's Test Prompt tab), without requiring the policy to
    // be attached to a registered model.
    register_post("routing/validate", [this](const httplib::Request& req, httplib::Response& res) {
        handle_routing_validate(req, res);
    });

    register_get("downloads", [this](const httplib::Request& req, httplib::Response& res) {
        handle_downloads(req, res);
    });

    register_post("downloads/control", [this](const httplib::Request& req, httplib::Response& res) {
        handle_download_control(req, res);
    });

    register_post("jobs", [this](const httplib::Request& req, httplib::Response& res) {
        handle_jobs_create(req, res);
    });
    register_get("jobs", [this](const httplib::Request& req, httplib::Response& res) {
        handle_jobs_list(req, res);
    });
    for (const char* prefix : {"/api/v0", "/api/v1", "/v0", "/v1"}) {
        web_server.Post(std::string(prefix) + R"(/jobs/([^/]+)/pause)",
                        [this](const httplib::Request& req, httplib::Response& res) {
                            handle_jobs_pause(req, res);
                        });
        web_server.Post(std::string(prefix) + R"(/jobs/([^/]+)/interrupt)",
                        [this](const httplib::Request& req, httplib::Response& res) {
                            handle_jobs_interrupt(req, res);
                        });
        web_server.Post(std::string(prefix) + R"(/jobs/([^/]+)/resume)",
                        [this](const httplib::Request& req, httplib::Response& res) {
                            handle_jobs_resume(req, res);
                        });
        web_server.Get(std::string(prefix) + R"(/jobs/([^/]+))",
                       [this](const httplib::Request& req, httplib::Response& res) {
                           handle_jobs_get(req, res);
                       });
        web_server.Delete(std::string(prefix) + R"(/jobs/([^/]+))",
                          [this](const httplib::Request& req, httplib::Response& res) {
                              handle_jobs_delete(req, res);
                          });
    }


    register_post("load", [this](const httplib::Request& req, httplib::Response& res) {
        handle_load(req, res);
    });

    register_post("load/command", [this](const httplib::Request& req, httplib::Response& res) {
        handle_load_command(req, res);
    });

    register_post("unload", [this](const httplib::Request& req, httplib::Response& res) {
        handle_unload(req, res);
    });

    register_post("delete", [this](const httplib::Request& req, httplib::Response& res) {
        handle_delete(req, res);
    });

    // Backend management endpoints
    register_post("install", [this](const httplib::Request& req, httplib::Response& res) {
        handle_install(req, res);
    });

    register_post("install/dry-run", [this](const httplib::Request& req, httplib::Response& res) {
        handle_install_dry_run(req, res);
    });

    register_post("uninstall", [this](const httplib::Request& req, httplib::Response& res) {
        handle_uninstall(req, res);
    });

    // System endpoints
    register_get("stats", [this](const httplib::Request& req, httplib::Response& res) {
        handle_stats(req, res);
    });

    register_get("system-info", [this](const httplib::Request& req, httplib::Response& res) {
        handle_system_info(req, res);
    });

    register_get("system-stats", [this](const httplib::Request& req, httplib::Response& res) {
        handle_system_stats(req, res);
    });


    // NOTE: /api/v1/halt endpoint removed - use SIGTERM signal instead (like Python server)
    // The stop command now sends termination signal directly to the process

    // Internal shutdown endpoint (not part of public API)
    web_server.Post("/internal/shutdown", [this](const httplib::Request& req, httplib::Response& res) {
        handle_shutdown(req, res);
    });

    web_server.Post("/internal/telemetry/flush", [](const httplib::Request& req, httplib::Response& res) {
        lemon::telemetry::flush();
        res.status = 200;
        res.set_content(nlohmann::json{{"status", "flushed"}}.dump(), "application/json");
    });

    web_server.Post("/internal/pin", [this](const httplib::Request& req, httplib::Response& res) {
        handle_pin(req, res);
    });

    // Unified config endpoints (not part of public API)
    web_server.Post("/internal/set", [this](const httplib::Request& req, httplib::Response& res) {
        handle_config_set(req, res);
    });
    web_server.Get("/internal/config", [this](const httplib::Request& req, httplib::Response& res) {
        handle_config_get(req, res);
    });
    web_server.Get("/internal/config/defaults", [this](const httplib::Request& req, httplib::Response& res) {
        handle_config_defaults_get(req, res);
    });
    web_server.Post("/internal/cleanup-cache", [this](const httplib::Request& req, httplib::Response& res) {
        handle_cleanup_cache(req, res);
    });
    web_server.Post("/internal/simulate-vram-pressure", [this](const httplib::Request& req, httplib::Response& res) {
        handle_simulate_vram_pressure(req, res);
    });
    web_server.Post("/internal/models/sync", [this](const httplib::Request& req, httplib::Response& res) {
        handle_models_sync(req, res);
    });
    web_server.Get("/internal/models/sync/status", [this](const httplib::Request& req, httplib::Response& res) {
        handle_models_sync_status(req, res);
    });

    web_server.Get("/internal/aliases", [this](const httplib::Request& req, httplib::Response& res) {
        handle_aliases_get(req, res);
    });
    web_server.Post("/internal/aliases", [this](const httplib::Request& req, httplib::Response& res) {
        handle_aliases_add(req, res);
    });
    web_server.Delete(R"(/internal/aliases/(.+))", [this](const httplib::Request& req, httplib::Response& res) {
        handle_aliases_remove(req, res);
    });

    // Server-side MCP client host foundation (admin-gated through the existing
    // /internal/* pre-routing auth). GUI3 and the web UI can both use these
    // endpoints via the normal Lemonade server connection.
    register_mcp_client_routes(web_server, cache_dir_, config_dir_);

    // Cloud auth: register quad-prefix POST and a parameterized DELETE.
    //   POST /v1/cloud/auth        body: {provider, api_key}
    //   DELETE /v1/cloud/auth/{p}
    // The runtime key lives in process memory only; env var
    // LEMONADE_<PROVIDER>_API_KEY takes precedence (POST returns 409 if it
    // is set). Both endpoints respect LEMONADE_ADMIN_API_KEY when configured
    // via the standard authentication path applied to /v1/.
    register_post("cloud/auth", [this](const httplib::Request& req, httplib::Response& res) {
        handle_cloud_auth_set(req, res);
    });
    web_server.Delete(R"(/api/v0/cloud/auth/(.+))", [this](const httplib::Request& req, httplib::Response& res) {
        handle_cloud_auth_clear(req, res);
    });
    web_server.Delete(R"(/api/v1/cloud/auth/(.+))", [this](const httplib::Request& req, httplib::Response& res) {
        handle_cloud_auth_clear(req, res);
    });
    web_server.Delete(R"(/v0/cloud/auth/(.+))", [this](const httplib::Request& req, httplib::Response& res) {
        handle_cloud_auth_clear(req, res);
    });
    web_server.Delete(R"(/v1/cloud/auth/(.+))", [this](const httplib::Request& req, httplib::Response& res) {
        handle_cloud_auth_clear(req, res);
    });

    // Register Ollama-compatible API routes
    auto ollama_api = std::make_shared<OllamaApi>(router_.get(), model_manager_.get());
    ollama_api->register_routes(web_server);

    // Register MCP gateway (POST /mcp). NOTE: /mcp is an INTENTIONAL EXCEPTION
    // to the quad-prefix invariant (AGENTS.md #1) — the MCP spec mandates a
    // single endpoint URL.
    auto mcp_server = std::make_shared<McpServer>(
        router_.get(),
        model_manager_.get(),
        [this](const std::string& m) { auto_load_model_if_needed(m); });
    mcp_server->register_routes(web_server);

    // Setup static file serving for web UI
    setup_static_files(web_server);
}

void Server::setup_static_files(httplib::Server &web_server) {
    // Determine static files directory (relative to executable)
    std::string static_dir = utils::get_resource_path("resources/static");

    // Create a reusable handler for serving index.html with template variable replacement
    auto serve_index_html = [this, static_dir](const httplib::Request&, httplib::Response& res) {
        std::string index_path = static_dir + "/index.html";
        std::ifstream file(index_path);

        if (!file.is_open()) {
            LOG(ERROR, "Server") << "Could not open index.html at: " << index_path << std::endl;
            res.status = 404;
            res.set_content("{\"error\": \"index.html not found\"}", "application/json");
            return;
        }

        // Read the entire file
        std::string html_template((std::istreambuf_iterator<char>(file)), std::istreambuf_iterator<char>());
        file.close();

        // Get filtered models from model manager
        auto models_map = model_manager_->get_supported_models();

        // Convert map to JSON
        json filtered_models = json::object();
        for (const auto& [model_name, info] : models_map) {
            std::vector<std::string> public_components;
            public_components.reserve(info.components.size());
            for (const auto& component : info.components) {
                public_components.push_back(model_manager_->get_public_model_name(component));
            }
            filtered_models[model_name] = {
                {"model_name", model_name},
                {"checkpoint", info.checkpoint()},
                {"recipe", info.recipe},
                {"labels", info.labels},
                {"suggested", info.suggested},
                {"source", info.source.empty() ? info.registry_source : info.source},
                {"registry_source", info.registry_source},
                {"components", public_components},
                {"mmproj", info.mmproj()}
            };

            // Add size if available
            if (info.size > 0.0) {
                filtered_models[model_name]["size"] = info.size;
            }
        }

        // Create JavaScript snippets
        std::string server_models_js = "<script>window.SERVER_MODELS = " + filtered_models.dump() + ";</script>";

        // Get platform name
        std::string platform_name;
        #ifdef _WIN32
            platform_name = "Windows";
        #elif __APPLE__
            platform_name = "Darwin";
        #elif __linux__
            platform_name = "Linux";
        #else
            platform_name = "Unknown";
        #endif
        std::string platform_js = "<script>window.PLATFORM = '" + platform_name + "';</script>";

        // Replace template variables
        size_t pos;

        // Replace {{SERVER_PORT}}
        while ((pos = html_template.find("{{SERVER_PORT}}")) != std::string::npos) {
            html_template.replace(pos, 15, std::to_string(port_));
        }

        // Replace {{SERVER_MODELS_JS}}
        while ((pos = html_template.find("{{SERVER_MODELS_JS}}")) != std::string::npos) {
            html_template.replace(pos, 20, server_models_js);
        }

        // Replace {{PLATFORM_JS}}
        while ((pos = html_template.find("{{PLATFORM_JS}}")) != std::string::npos) {
            html_template.replace(pos, 15, platform_js);
        }

        // Set no-cache headers
        res.set_header("Cache-Control", "no-cache, no-store, must-revalidate");
        res.set_header("Pragma", "no-cache");
        res.set_header("Expires", "0");
        res.set_content(html_template, "text/html");
    };

    // Serve index.html at /api/v1 for compatibility
    web_server.Get("/api/v1", serve_index_html);

    // Mount static files directory for status page assets (CSS, JS, images)
    if (!web_server.set_mount_point("/static", static_dir)) {
        LOG(WARNING, "Server") << "Could not mount static files from: " << static_dir << std::endl;
        LOG(WARNING, "Server") << "Status page assets will not be available" << std::endl;
    }

    // Web app UI endpoint - serve the React web app at root
    std::string web_app_dir = utils::get_resource_path("resources/web-app");

    // Check if web app directory exists
    if (fs::exists(web_app_dir) && fs::is_directory(web_app_dir)) {
        // Create a handler for serving web app index.html for SPA routing
        auto serve_web_app_html = [web_app_dir](const httplib::Request&, httplib::Response& res) {
            std::string index_path = web_app_dir + "/index.html";
            std::ifstream file(index_path);

            if (!file.is_open()) {
                res.status = 404;
                res.set_content("{\"error\": \"Web app not found\"}", "application/json");
                return;
            }

            std::string html((std::istreambuf_iterator<char>(file)), std::istreambuf_iterator<char>());
            file.close();

            // Inject mock window.api for web compatibility with the shared Tauri app renderer
            std::string mock_api = R"(
<script>
// Mock window.api for web compatibility (the Tauri shim is skipped in pure-web mode)
window.api = {
    isWebApp: true,  // Explicit flag to indicate web mode
    platform: navigator.platform || 'web',
    minimizeWindow: () => {},
    maximizeWindow: () => {},
    closeWindow: () => {},
    openExternal: (url) => window.open(url, '_blank'),
    onMaximizeChange: () => {},
    updateMinWidth: () => {},
    zoomIn: () => document.body.style.zoom = (parseFloat(document.body.style.zoom || '1') + 0.1).toString(),
    zoomOut: () => document.body.style.zoom = (parseFloat(document.body.style.zoom || '1') - 0.1).toString(),
    getSettings: async () => {
        const saved = localStorage.getItem('lemonade-settings');
        if (saved) return JSON.parse(saved);
        // Return defaults matching DEFAULT_LAYOUT_SETTINGS from appSettings.ts
        return {
            layout: {
                isChatVisible: true,
                isModelManagerVisible: true,
                isCenterPanelVisible: true,
                isLogsVisible: false,
                modelManagerWidth: 280,
                chatWidth: 350,
                logsHeight: 200
            },
            theme: 'dark',
            apiUrl: window.location.origin,
            apiKey: { value: '' }
        };
    },
    saveSettings: async (settings) => {
        localStorage.setItem('lemonade-settings', JSON.stringify(settings));
        return settings;
    },
    onSettingsUpdated: () => {},
    getServerPort: () => parseInt(window.location.port) || 13305,
    onServerPortUpdated: () => {},
    getServerAPIKey: async () => {
        const settings = await window.api.getSettings();
        return settings.apiKey?.value || '';
    },
    restartApp: () => window.location.reload(),
    writeClipboard: async (text) => {
        if (navigator.clipboard) {
            try {
                await navigator.clipboard.writeText(text);
                return;
            } catch {
                // Ignore clipboard errors and fall back to legacy method
            }
        }
        const ta = document.createElement('textarea');
        ta.value = String(text);
        ta.setAttribute('readonly', '');
        ta.style.position = 'fixed';
        ta.style.left = '-9999px';
        document.body.appendChild(ta);
        try {
            ta.select();
            if (!document.execCommand('copy')) {
                throw new Error('Legacy clipboard copy failed');
            }
        } finally {
            document.body.removeChild(ta);
        }
    }
};
</script>
)";

            // Insert mock API before the closing </head> tag
            size_t head_end_pos = html.find("</head>");
            if (head_end_pos != std::string::npos) {
                html.insert(head_end_pos, mock_api);
            }

            // Set no-cache headers
            res.set_header("Cache-Control", "no-cache, no-store, must-revalidate");
            res.set_header("Pragma", "no-cache");
            res.set_header("Expires", "0");
            res.set_content(html, "text/html");
        };

        // Serve the web app's index.html at root and for SPA routes
        web_server.Get("/", serve_web_app_html);

        // Also serve at /web-app for backwards compatibility
        web_server.Get("/web-app/?", serve_web_app_html);

        // Serve all static assets from the web app directory (JS, CSS, fonts, assets, etc.)
        // Handle both root-level assets and /web-app/ prefixed paths for backwards compatibility
        auto serve_web_app_asset = [web_app_dir](const httplib::Request& req, httplib::Response& res, const std::string& file_path) {
            std::error_code ec;
            namespace fs = std::filesystem;

            auto base = fs::weakly_canonical(fs::path(web_app_dir), ec);
            if (ec) {
                res.status = 500;
                res.set_content("Internal server error", "text/plain");
                return;
            }

            auto candidate = fs::weakly_canonical(base / file_path, ec);
            if (ec) {
                // Path doesn't exist or cannot be canonicalized
                res.status = 404;
                res.set_content("File not found", "text/plain");
                return;
            }

            // Verify the resolved path is confined under the base directory.
            // Use std::filesystem::relative (not string prefix) so it works
            // correctly on Windows where path separators differ.
            auto relative = fs::relative(candidate, base, ec);
            if (ec || relative.empty() || relative.is_absolute()) {
                // empty = candidate == base (directory, not a file)
                // is_absolute = somehow escaped (shouldn't happen after canonicalization)
                res.status = 403;
                res.set_content("Forbidden", "text/plain");
                return;
            }
            // Belt-and-suspenders: reject if any path component is ".."
            // (weakly_canonical should have resolved it, but this catches
            // edge cases on exotic filesystems). Check components, not
            // substrings, so legitimate filenames like "my..file.js" are allowed.
            for (const auto& part : relative) {
                if (part == "..") {
                    res.status = 403;
                    res.set_content("Forbidden", "text/plain");
                    return;
                }
            }

            // Serve the file
            std::ifstream file(candidate, std::ios::binary);
            if (!file.is_open()) {
                res.status = 404;
                res.set_content("File not found", "text/plain");
                return;
            }

            // Read file content
            std::string content((std::istreambuf_iterator<char>(file)), std::istreambuf_iterator<char>());
            file.close();

            // Determine content type based on extension
            std::string content_type = "application/octet-stream";
            size_t dot_pos = file_path.rfind('.');
            if (dot_pos != std::string::npos) {
                std::string ext = file_path.substr(dot_pos);
                if (ext == ".js") content_type = "text/javascript";
                else if (ext == ".css") content_type = "text/css";
                else if (ext == ".html") content_type = "text/html";
                else if (ext == ".woff") content_type = "font/woff";
                else if (ext == ".woff2") content_type = "font/woff2";
                else if (ext == ".ttf") content_type = "font/ttf";
                else if (ext == ".svg") content_type = "image/svg+xml";
                else if (ext == ".png") content_type = "image/png";
                else if (ext == ".jpg" || ext == ".jpeg") content_type = "image/jpeg";
                else if (ext == ".json") content_type = "application/json";
                else if (ext == ".ico") content_type = "image/x-icon";
            }

            res.set_content(content, content_type);
        };

        // Serve favicon from web-app directory at root
        web_server.Get("/favicon.ico", [serve_web_app_asset](const httplib::Request& req, httplib::Response& res) {
            serve_web_app_asset(req, res, "favicon.ico");
        });

        // Serve web app assets from root (for files like renderer.bundle.js, fonts, etc.)
        web_server.Get(R"(/([^/]+\.(js|css|woff|woff2|ttf|svg|png|jpg|jpeg|json|ico)))",
                      [serve_web_app_asset](const httplib::Request& req, httplib::Response& res) {
            std::string file_path = req.matches[1].str();
            serve_web_app_asset(req, res, file_path);
        });

        // Keep /web-app/ prefix routes for backwards compatibility
        web_server.Get(R"(/web-app/(.+))", [serve_web_app_asset](const httplib::Request& req, httplib::Response& res) {
            std::string file_path = req.matches[1].str();
            serve_web_app_asset(req, res, file_path);
        });

        // SPA fallback: serve index.html for any unmatched GET routes that don't start with /api, /v0, /v1, /static, or /live
        // This enables client-side routing
        web_server.Get(R"(^(?!/api|/v0|/v1|/static|/live|/status|/internal).*)",
                      [serve_web_app_html](const httplib::Request& req, httplib::Response& res) {
            // Only serve index.html if the path doesn't look like a file with extension
            std::string path = req.path;
            size_t last_slash = path.rfind('/');
            std::string last_segment = (last_slash != std::string::npos) ? path.substr(last_slash + 1) : path;

            // If the last segment has an extension and it's not .html, let it 404
            // (This helps catch missing assets more clearly)
            size_t dot_pos = last_segment.rfind('.');
            if (dot_pos != std::string::npos) {
                std::string ext = last_segment.substr(dot_pos);
                if (ext != ".html" && ext != ".htm") {
                    // File with extension not found, return 404
                    res.status = 404;
                    return;
                }
            }

            // Otherwise, serve the SPA index.html for client-side routing
            serve_web_app_html(req, res);
        });
    } else {
        // Fallback to static page when web-app is not compiled
        LOG(INFO, "Server") << "Web app directory not found at: " << web_app_dir << std::endl;
        LOG(INFO, "Server") << "Falling back to static status page at root" << std::endl;

        // Serve the static status page at root instead
        web_server.Get("/", serve_index_html);

        // Serve favicon from static directory
        web_server.Get("/favicon.ico", [static_dir](const httplib::Request& req, httplib::Response& res) {
            std::ifstream ifs(static_dir + "/favicon.ico", std::ios::binary);
            if (ifs) {
                std::string content((std::istreambuf_iterator<char>(ifs)), (std::istreambuf_iterator<char>()));
                res.set_content(content, "image/x-icon");
                res.status = 200;
            } else {
                res.set_content("Favicon not found.", "text/plain");
                res.status = 404;
            }
        });
    }

    // Override default headers for static files to include no-cache
    // This ensures the web UI always gets the latest version
    web_server.set_file_request_handler([](const httplib::Request& req, httplib::Response& res) {
        // Add no-cache headers for static files
        res.set_header("Cache-Control", "no-cache, no-store, must-revalidate");
        res.set_header("Pragma", "no-cache");
        res.set_header("Expires", "0");
    });
}

void Server::setup_cors(httplib::Server &web_server) {
    // Set CORS headers for all responses
    web_server.set_default_headers({
        {"Access-Control-Allow-Methods", "GET, POST, PUT, DELETE, OPTIONS"},
        {"Access-Control-Allow-Headers", "Content-Type, Authorization, X-Client-Session-Id, X-Account-Session-Id, mcp-protocol-version, traceparent, Mcp-Session-Id"}
    });

    // Handle preflight OPTIONS requests
    web_server.Options(".*", [](const httplib::Request&, httplib::Response& res) {
        res.status = 204;
    });

    // Catch-all error handler - must be last!
    web_server.set_error_handler([](const httplib::Request& req, httplib::Response& res) {
        LOG(ERROR, "Server") << "Error " << res.status << ": " << req.method << " " << req.path << std::endl;

        if (res.status == 404) {
            // Only set generic "endpoint not found" if no content was already set
            // This preserves specific error messages (e.g., "model not found")
            if (res.body.empty()) {
                nlohmann::json error = {
                    {"error", {
                        {"message", "The requested endpoint does not exist"},
                        {"type", "not_found"},
                        {"path", req.path}
                    }}
                };
                res.set_content(error.dump(), "application/json");
            }
        } else if (res.status == 400) {
            // Log more details about 400 errors
            LOG(ERROR, "Server") << "400 Bad Request details - Body length: " << req.body.length()
                      << ", Content-Type: " << req.get_header_value("Content-Type") << std::endl;
            // Ensure a response is sent
            if (res.body.empty()) {
                nlohmann::json error = {
                    {"error", {
                        {"message", "Bad request"},
                        {"type", "bad_request"}
                    }}
                };
                res.set_content(error.dump(), "application/json");
            }
        }
    });
}

std::string Server::resolve_host_to_ip(int ai_family, const std::string& host) {
    struct addrinfo hints = {0};
    hints.ai_family = ai_family;
    hints.ai_socktype = SOCK_STREAM;
    hints.ai_flags = 0; // No AI_ADDRCONFIG: allows loopback resolution when offline

    struct addrinfo *result = nullptr;

    // Check return value (0 is success)
    if (getaddrinfo(host.c_str(), nullptr, &hints, &result) != 0) {
        LOG(WARNING, "Server") << "resolution failed for " << host << " no " << (ai_family == AF_INET ? "IPv4" : ai_family == AF_INET6 ? "IPv6" : "") << " resolution found." << std::endl;
        return ""; // Return empty string on failure, don't return void
    }

    if (result == nullptr) return "";

    // Use INET6_ADDRSTRLEN to be safe for both (it's larger)
    char addrstr[INET6_ADDRSTRLEN];
    void *ptr = nullptr;

    // Safety Check - verify what we actually got back
    if (result->ai_family == AF_INET) {
        struct sockaddr_in *ipv4 = (struct sockaddr_in *)result->ai_addr;
        ptr = &(ipv4->sin_addr);
    } else if (result->ai_family == AF_INET6) {
        struct sockaddr_in6 *ipv6 = (struct sockaddr_in6 *)result->ai_addr;
        ptr = &(ipv6->sin6_addr);
    } else {
        freeaddrinfo(result);
        return "";
    }

    // Convert binary IP to string
    inet_ntop(result->ai_family, ptr, addrstr, sizeof(addrstr));

    std::string resolved_ip(addrstr);
    freeaddrinfo(result);
    return resolved_ip;
}

void Server::setup_http_logger(httplib::Server &web_server) {
    // Add request logging for ALL requests (except health checks and stats endpoints)
    web_server.set_logger([this](const httplib::Request& req, const httplib::Response& res) {
        if (req.path == "/metrics") {
            if (res.status == 200) {
                bool expected = false;
                if (metrics_access_logged_.compare_exchange_strong(expected, true)) {
                    LOG(INFO, "Server") << req.method << " " << req.path << " - " << res.status << std::endl;
                }
            } else {
                LOG(WARNING, "Server") << req.method << " " << req.path << " - " << res.status << std::endl;
            }
            return;
        }

        // Skip logging health checks and stats endpoints to reduce log noise
        if (req.path == "/api/v0/health" || req.path == "/api/v1/health" ||
            req.path == "/v0/health" || req.path == "/v1/health" || req.path == "/live" ||
            is_quiet_polling_path(req.path)) {
            return;
        }

        // Determine if this is a high-signal request or noise (static assets, repeated queries)
        bool is_quiet_get = (req.method == "GET" && (
            req.path == "/" ||
            req.path == "/api/v0/models" || req.path == "/api/v1/models" ||
            req.path == "/v0/models" || req.path == "/v1/models" ||
            req.path == "/api/v0/system-info" || req.path == "/api/v1/system-info" ||
            req.path == "/v0/system-info" || req.path == "/v1/system-info" ||
            req.path == "/api/v0/system-checks" || req.path == "/api/v1/system-checks" ||
            req.path == "/v0/system-checks" || req.path == "/v1/system-checks" ||
            req.path.find(".js") != std::string::npos ||
            req.path.find(".css") != std::string::npos ||
            req.path.find(".svg") != std::string::npos ||
            req.path.find(".png") != std::string::npos ||
            req.path.find(".ico") != std::string::npos ||
            req.path.find(".woff") != std::string::npos
        ));

        if (!is_quiet_get) {
            LOG(DEBUG, "Server") << req.method << " " << req.path << " - " << res.status << std::endl;
        }
    });
}

void Server::run() {
    std::string host = config_->host();
    LOG(INFO, "Server") << "Starting HTTP server on " << host << ":" << port_ << std::endl;

    std::string ipv4 = resolve_host_to_ip(AF_INET, host);
    std::string ipv6 = resolve_host_to_ip(AF_INET6, host);

    LOG(INFO, "Server") << "Host resolution: IPv4=" << (ipv4.empty() ? "(none)" : ipv4)
                        << ", IPv6=" << (ipv6.empty() ? "(none)" : ipv6) << std::endl;

    if (ipv4.empty() && ipv6.empty()) {
        throw std::runtime_error("Failed to resolve host '" + host + "' to any address. "
                                 "Cannot start server.");
    }

    // Fail fast if the port is already taken (usually another lemond). Detecting
    // it here keeps the error from being buried under later startup logs.
    {
        std::string in_use_ip;
        if (!ipv4.empty() && utils::is_tcp_listener_active(AF_INET, ipv4, port_)) {
            in_use_ip = ipv4;
        } else if (!ipv6.empty() && utils::is_tcp_listener_active(AF_INET6, ipv6, port_)) {
            in_use_ip = ipv6;
        }
        if (!in_use_ip.empty()) {
            std::string msg = "Port " + std::to_string(port_) + " on " + in_use_ip +
                " is already in use. Another Lemonade server (lemond) is likely "
                "already running on this port. This instance will now exit.";
            std::cerr << "[Server] ERROR: " << msg << std::endl;  // terminal visibility
            LOG(ERROR, "Server") << msg << std::endl;
            startup_failed_ = true;
            return;
        }
    }

    // Operators binding beyond loopback should secure the server with an API
    // key, since every endpoint is reachable from other machines once the host
    // is non-loopback. The regular API routes (/api, /v0, /v1) are gated by
    // api_key_; the /internal/* control endpoints (shutdown, set, config) are
    // gated by admin_api_key_, which defaults to api_key_. Setting only
    // LEMONADE_ADMIN_API_KEY therefore protects /internal/* but still leaves the
    // inference and model-management endpoints exposed, so we warn unless the
    // regular key is set.
    auto warn_if_unsecured = [this](const std::string& bound_host,
                                    const std::string& v4, const std::string& v6) {
        auto is_loopback = [](const std::string& ip) {
            return ip.empty() || ip.rfind("127.", 0) == 0 || ip == "::1";
        };
        if (is_loopback(v4) && is_loopback(v6)) {
            return;
        }
        if (api_key_.empty() && admin_api_key_.empty()) {
            LOG(WARNING, "Server")
                << "Serving on non-loopback host '" << bound_host
                << "' without an API key. All endpoints, including the /internal/* "
                   "control endpoints and the /internal/mcp/* process-launch endpoints, "
                   "are reachable from other machines unauthenticated. Set "
                   "LEMONADE_API_KEY to secure all endpoints; LEMONADE_ADMIN_API_KEY "
                   "on its own only secures the /internal/* "
                   "control endpoints." << std::endl;
        } else if (api_key_.empty()) {
            LOG(WARNING, "Server")
                << "Serving on non-loopback host '" << bound_host
                << "' with only an admin API key set. The /internal/* control "
                   "endpoints are protected, but the inference and model-management "
                   "endpoints (/api, /v0, /v1) are reachable from other machines "
                   "unauthenticated. Set LEMONADE_API_KEY to secure them." << std::endl;
        }
    };
    warn_if_unsecured(host, ipv4, ipv6);

    running_ = true;

    // Start WebSocket server for realtime API and log streaming
    if (websocket_server_) {
        if (websocket_server_->start()) {
            LOG(INFO, "Server") << "WebSocket server started on port "
                                << websocket_server_->get_port() << std::endl;
        } else {
            LOG(WARNING, "Server") << "Failed to start WebSocket server" << std::endl;
        }
    }

    while (true) {
        // Check for shutdown signal from the main thread
        if (shutdown_requested_.load()) {
            LOG(INFO, "Server") << "Shutdown requested, stopping server..." << std::endl;
            stop();
            break;
        }

        const std::size_t listener_count =
            static_cast<std::size_t>(!ipv4.empty()) +
            static_cast<std::size_t>(!ipv6.empty());
        utils::ListenerStartupState listener_startup(listener_count);

        if (!ipv4.empty()) {
            // setup ipv4 thread
            setup_http_logger(*http_server_);
            http_v4_thread_ = std::thread([this, ipv4, &listener_startup]() {
                LOG(INFO, "Server") << "Binding IPv4 HTTP server to " << ipv4 << ":" << port_ << "..." << std::endl;
                int result = http_front_->bind_to_port(ipv4, port_);
                if (result <= 0) {
                    LOG(ERROR, "Server") << "Failed to bind IPv4 HTTP server to " << ipv4 << ":" << port_ << std::endl;
                    listener_startup.record_bind_failure();
                    return;
                }
                // The routed server's keep-alive loop runs only while it sees
                // a valid listen socket
                http_server_->set_listen_socket(http_front_->listen_socket());
                LOG(INFO, "Server") << "IPv4 HTTP server listening on " << ipv4 << ":" << port_ << std::endl;
                listener_startup.record_bind_success();
                if (!http_front_->listen_after_bind()) {
                    LOG(ERROR, "Server") << "IPv4 HTTP server listen_after_bind() failed" << std::endl;
                    listener_startup.record_listen_failure();
                }
            });
        }
        if (!ipv6.empty()) {
            // setup ipv6 thread
            setup_http_logger(*http_server_v6_);
            http_v6_thread_ = std::thread([this, ipv6, &listener_startup]() {
                LOG(INFO, "Server") << "Binding IPv6 HTTP server to [" << ipv6 << "]:" << port_ << "..." << std::endl;
                int result = http_front_v6_->bind_to_port(ipv6, port_);
                if (result <= 0) {
                    LOG(ERROR, "Server") << "Failed to bind IPv6 HTTP server to [" << ipv6 << "]:" << port_ << std::endl;
                    listener_startup.record_bind_failure();
                    return;
                }
                http_server_v6_->set_listen_socket(http_front_v6_->listen_socket());
                LOG(INFO, "Server") << "IPv6 HTTP server listening on [" << ipv6 << "]:" << port_ << std::endl;
                listener_startup.record_bind_success();
                if (!http_front_v6_->listen_after_bind()) {
                    LOG(ERROR, "Server") << "IPv6 HTTP server listen_after_bind() failed" << std::endl;
                    listener_startup.record_listen_failure();
                }
            });
        }

        while (!listener_startup.bind_attempts_complete() &&
               !shutdown_requested_.load() && !rebind_requested_.load()) {
            std::this_thread::sleep_for(std::chrono::milliseconds(10));
        }

        if (listener_startup.failed()) {
            LOG(ERROR, "Server") << "Could not start HTTP listeners for every address resolved for host '"
                                 << host << "' (bind or listen failure). Server startup aborted." << std::endl;
            stop();
            if (http_v4_thread_.joinable())
                http_v4_thread_.join();
            if (http_v6_thread_.joinable())
                http_v6_thread_.join();
            startup_failed_ = true;
            break;
        }

        // Enumerate all RFC1918 interfaces to determine if we can broadcast.
        // The beacon will send per-interface with the correct IP in the payload.
        auto rfc1918Interfaces = udp_beacon_.getLocalRFC1918Interfaces();
        bool bcast = config_->broadcast();
        if (!rfc1918Interfaces.empty() && bcast) {
            std::cout << "[Server] [Net Broadcast] Broadcasting on " << rfc1918Interfaces.size()
                      << " RFC1918 interface(s):";
            for (const auto& iface : rfc1918Interfaces) {
                std::cout << " " << iface.ipAddress << " (bcast " << iface.broadcastAddress << ")";
            }
            std::cout << std::endl;
            udp_beacon_.startBroadcasting(
                13305, // Broadcast port best to not make it adjustable, so clients dont have to scan.
                port_,
                2
            );
        } else if (!rfc1918Interfaces.empty() && !bcast) {
            LOG(INFO, "Server") << "Broadcasting disabled by configuration or CLI option" << std::endl;
        } else {
            LOG(INFO, "Server") << "Unable to broadcast my existance please use a RFC1918 IPv4," << std::endl
                        << "or hostname that resolves to RFC1918 IPv4." << std::endl;
        }

        // Wait for listener threads, but check periodically for shutdown or rebind signals.
        // The threads are blocked in listen_after_bind(), which only returns when
        // the server is stopped or an error occurs.
        while ((http_v4_thread_.joinable() || http_v6_thread_.joinable()) &&
               !shutdown_requested_.load() && !rebind_requested_.load() &&
               !listener_startup.failed()) {
            std::this_thread::sleep_for(std::chrono::milliseconds(100));
        }

        // If shutdown was requested while the server was running, stop it now
        // to unblock the listener threads (they're stuck in listen_after_bind)
        if (shutdown_requested_.load()) {
            LOG(INFO, "Server") << "Shutdown requested, stopping server..." << std::endl;
            stop();
            // Join the threads (they should exit quickly after stop() is called)
            if (http_v4_thread_.joinable())
                http_v4_thread_.join();
            if (http_v6_thread_.joinable())
                http_v6_thread_.join();
            break;  // Exit the main loop
        }

        if (listener_startup.failed()) {
            LOG(ERROR, "Server") << "An HTTP listener stopped unexpectedly. Server exiting." << std::endl;
            stop();
            if (http_v4_thread_.joinable())
                http_v4_thread_.join();
            if (http_v6_thread_.joinable())
                http_v6_thread_.join();
            startup_failed_ = true;
            break;
        }

        // If rebind was requested, stop() has already been called by apply_config_side_effects().
        // Just join the threads so they can be restarted with new settings.
        if (rebind_requested_.load()) {
            // Wait for threads to finish (stop() was already called)
            if (http_v4_thread_.joinable())
                http_v4_thread_.join();
            if (http_v6_thread_.joinable())
                http_v6_thread_.join();
            // Continue to rebind logic below (don't break)
        } else {
            // Normal path: threads exited naturally (no shutdown, no rebind)
            // Join the threads
            if (http_v4_thread_.joinable())
                http_v4_thread_.join();
            if (http_v6_thread_.joinable())
                http_v6_thread_.join();
        }

        if (!rebind_requested_) {
            break;  // Normal exit (stop() was called)
        }

        // Rebind requested: re-resolve host, recreate HTTP servers, loop back to bind+listen
        host = config_->host();
        ipv4 = resolve_host_to_ip(AF_INET, host);
        ipv6 = resolve_host_to_ip(AF_INET6, host);
        warn_if_unsecured(host, ipv4, ipv6);
        LOG(INFO, "Server") << "Rebinding to " << host << ":" << port_ << "..." << std::endl;
        rebind_requested_ = false;
        setup_http_servers();
    }
}


bool Server::should_shutdown() const {
    return shutdown_requested_.load();
}

void Server::set_shutdown_requested(bool requested) {
    shutdown_requested_.store(requested);
}

bool Server::is_running() const {
    return running_;
}

bool Server::startup_failed() const {
    return startup_failed_;
}

void Server::stop() {
    if (running_) {
        LOG(INFO, "Server") << "Stopping HTTP server..." << std::endl;
        udp_beacon_.stopBroadcasting();
        stop_http_listeners();
        running_ = false;
        shutdown_requested_ = false;  // Reset for potential future use

        if (websocket_server_) {
            LOG(INFO, "Server") << "Stopping WebSocket server..." << std::endl;
            websocket_server_->stop();
        }

        if (router_) {
            LOG(INFO, "Server") << "Unloading models and stopping backend servers..." << std::endl;
            try {
                router_->unload_model();
            } catch (const std::exception& e) {
                LOG(ERROR, "Server") << "Error during cleanup: " << e.what() << std::endl;
            }
        }

        LOG(INFO, "Server") << "Shutting down telemetry queue..." << std::endl;
        lemon::telemetry::shutdown();

        LOG(INFO, "Server") << "Cleanup complete" << std::endl;
    }

    if (model_manager_) {
        model_manager_->cancel_sync();
    }

    if (model_cache_warmup_thread_.joinable()) {
        model_cache_warmup_thread_.join();
    }

    {
        std::lock_guard<std::mutex> lock(background_sync_mutex_);
        for (auto& task : background_sync_threads_) {
            if (task.thread.joinable()) {
                task.thread.join();
            }
        }
        background_sync_threads_.clear();
    }
}

// Generates an actionable error message for model loading failures.
// Handles three cases:
//   1. Model exists but was filtered out (e.g., NPU model on non-NPU system)
//   2. Model doesn't exist in the registry at all
//   3. Model exists but failed to load (engine error)
nlohmann::json Server::create_model_error(const std::string& requested_model, const std::string& exception_msg) {
    nlohmann::json error_response;

    // Case 1: Check if this model exists but was filtered out due to system requirements
    std::string filter_reason = model_manager_->get_model_filter_reason(requested_model);

    if (!filter_reason.empty()) {
        // Model exists but is not available on this system
        std::string message = "Model '" + requested_model + "' is not available on this system. " + filter_reason;

        error_response["error"] = {
            {"message", message},
            {"type", "model_not_supported"},
            {"param", "model"},
            {"code", "model_not_supported"},
            {"requested_model", requested_model}
        };

        return error_response;
    }

    // Case 2: Check if model doesn't exist in the registry at all
    if (!model_manager_->model_exists(requested_model)) {
        std::string message = "Model '" + requested_model + "' was not found. ";

        // Get available models and suggest some
        auto available_models = model_manager_->get_supported_models();

        if (!available_models.empty()) {
            // Collect model names
            std::vector<std::string> model_names;
            model_names.reserve(available_models.size());
            for (const auto& [name, info] : available_models) {
                model_names.push_back(name);
            }

            // Sort alphabetically for consistent output
            std::sort(model_names.begin(), model_names.end());

            // Show up to 3 available models
            const size_t max_suggestions = 3;
            size_t count = std::min(model_names.size(), max_suggestions);

            message += "Available models include: ";
            for (size_t i = 0; i < count; ++i) {
                if (i > 0) message += ", ";
                message += "'" + model_names[i] + "'";
            }

            if (model_names.size() > max_suggestions) {
                message += ", and " + std::to_string(model_names.size() - max_suggestions) + " more";
            }
            message += ". ";
        }

        message += "Use 'lemonade list' or GET /api/v1/models?show_all=true to see all available models.";

        // Add FLM hint for -FLM model names when FLM is not ready
        if (requested_model.size() > 4 &&
            requested_model.substr(requested_model.size() - 4) == "-FLM") {
            auto flm_status = SystemInfoCache::get_flm_status();
            if (!flm_status.is_ready()) {
                message += " The FLM backend is not ready: " + flm_status.message + ".";
                if (!flm_status.action.empty()) {
                    message += " " + flm_status.action + ".";
                }
            }
        }

        error_response["error"] = {
            {"message", message},
            {"type", "model_not_found"},
            {"param", "model"},
            {"code", "model_not_found"},
            {"requested_model", requested_model}
        };

        return error_response;
    }

    // Case 3: Model exists and is available, but failed to load.
    if (exception_msg.rfind("Routing residency conflict:", 0) == 0) {
        error_response["error"] = {
            {"message", exception_msg},
            {"type", "router_residency_conflict"},
            {"param", "model"},
            {"code", "router_residency_conflict"},
            {"requested_model", requested_model}
        };
        return error_response;
    }

    if (exception_msg.find("are pinned") != std::string::npos) {
        error_response["error"] = {
            {"message", exception_msg},
            {"type", "slots_pinned_error"},
            {"param", "model"},
            {"code", "slots_pinned_error"},
            {"requested_model", requested_model}
        };
        return error_response;
    }

    // Return the actual exception message so the user knows what went wrong
    std::string message = "Failed to load model '" + requested_model + "': " + exception_msg;

    error_response["error"] = {
        {"message", message},
        {"type", "model_load_error"},
        {"param", "model"},
        {"code", "model_load_error"},
        {"requested_model", requested_model}
    };

    return error_response;
}

// This function is called by:
//   - handle_chat_completions() - /chat/completions endpoint
//   - handle_completions() - /completions endpoint
//   - handle_load() - /load endpoint
//
// Behavior:
//   1. If model is already loaded: Return immediately (no-op)
//   2. If model is not downloaded: Download it (first-time use)
//   3. If model is downloaded: Use cached version
//      (do not check the remote registry for updates)
//
// Note: Only the /pull endpoint checks the model's recorded registry for
// updates (do_not_upgrade=false).

// Load-level options that may be forwarded to RecipeOptions during auto-load.
// Keep this an explicit allowlist so request-scoped fields do not leak into
// recipe options.
nlohmann::json Server::extract_auto_load_options(const json& request) {
    nlohmann::json result = json::object();

    auto extract_if_present =
        [&request, &result](const std::string& key) {
            if (request.contains(key)) {
                result[key] = request[key];
            }
        };

    extract_if_present("ctx_size");

    return result;
}

void Server::auto_load_model_if_needed(
    const std::string& requested_model,
    const json& request_options,
    LoadPurpose load_purpose) {
    // A live process follows its current use without a reload: routing work
    // promotes it to RoutingHelper, while direct inference demotes it into the
    // counted Standard pool. Destination-pool admission remains authoritative.
    if (router_->ensure_loaded_model_residency(
            requested_model, load_purpose)) {
        LOG(DEBUG, "Server")
            << "Model already loaded: " << requested_model
            << " (residency="
            << residency_class_to_string(
                   residency_class_for_load_purpose(load_purpose))
            << ")" << std::endl;
        if (request_options.contains("ctx_size")) {
            auto loaded_ctx = router_->get_model_recipe_options(requested_model)
                                  .get_option("ctx_size");
            LOG(DEBUG, "Server")
                << "Ignoring requested ctx_size=" << request_options["ctx_size"]
                << " for already-loaded " << requested_model
                << " (loaded ctx_size=" << loaded_ctx << ")" << std::endl;
        }
        return;
    }

    // Log the auto-loading action
    LOG(INFO, "Server") << "Auto-loading model: " << requested_model << std::endl;

    // Get model info
    if (!model_manager_->model_exists(requested_model)) {
        throw std::runtime_error("Model not found: " + requested_model);
    }

    auto info = model_manager_->get_model_info(requested_model);

    // Collections have no backend of their own — load each component instead.
    if (is_omni_collection_recipe(info.recipe)) {
        ensure_collection_loaded(info);
        return;
    }

    // Download model if not cached (first-time use)
    // IMPORTANT: Use do_not_upgrade=true to prevent checking the remote registry for updates
    // This means:
    //   - If model is NOT downloaded: Download it from its recorded registry
    //   - If model IS downloaded: Skip the registry API check entirely (use cached version)
    // Only the /pull endpoint should check for updates (uses do_not_upgrade=false)
    if (!model_manager_->backend_self_manages_downloads(info.recipe) &&
        !model_manager_->is_model_downloaded(requested_model)) {
        LOG(INFO, "Server") << "Model not cached, downloading from "
                            << remote_registry_display_name(
                                   parse_remote_registry_source(info.registry_source))
                            << "..." << std::endl;
        LOG(INFO, "Server") << "This may take several minutes for large models." << std::endl;
        model_manager_->download_registered_model(info, true);
        LOG(INFO, "Server") << "Model download complete: " << requested_model << std::endl;

        // CRITICAL: Refresh model info after download to get correct resolved_path
        // The resolved_path is computed based on filesystem, so we need fresh info now that files exist
        info = model_manager_->get_model_info(requested_model);
    }

    // Load model with do_not_upgrade=true, applying per-request options on first load.
    // For FLM models: FastFlowLMServer will handle download internally if needed
    // For non-FLM models: Model should already be cached at this point
    router_->load_model(requested_model, info,
                        RecipeOptions(info.recipe, request_options), true,
                        /*allow_reload_on_option_change=*/false,
                        /*pinned=*/std::nullopt,
                        load_purpose);
    LOG(INFO, "Server") << "Model loaded successfully: " << requested_model << std::endl;
}

void Server::ensure_collection_loaded(const ModelInfo& info) {
    LOG(INFO, "Server") << "Loading collection components for: " << info.model_name << std::endl;
    for (const auto& component : info.components) {
        if (!model_manager_->model_exists(component)) {
            LOG(WARNING, "Server") << "Skipping unknown component: " << component << std::endl;
            continue;
        }
        if (router_->is_model_loaded(component)) {
            LOG(DEBUG, "Server") << "Component already loaded: " << component << std::endl;
            continue;
        }
        auto comp_info = model_manager_->get_model_info(component);
        if (!comp_info.downloaded) {
            LOG(INFO, "Server") << "Downloading component: " << component << std::endl;
            model_manager_->download_registered_model(comp_info);
            comp_info = model_manager_->get_model_info(component);
        }
        LOG(INFO, "Server") << "Loading component: " << component << std::endl;
        // The collection load request does not become the component request
        // layer. Router already reads the component's saved options from comp_info,
        // so an empty request preserves normal model/architecture/backend scope
        // semantics.
        router_->load_model(
            component, comp_info, RecipeOptions(comp_info.recipe, json::object()), true,
            /*allow_reload_on_option_change=*/true);
    }
}

void Server::handle_health(const httplib::Request& req, httplib::Response& res) {
    // For HEAD requests, just return 200 OK without processing
    if (req.method == "HEAD") {
        res.status = 200;
        return;
    }

    nlohmann::json response = {{"status", "ok"}};

    // Add version information
    response["version"] = LEMON_VERSION_STRING;

    // Add telemetry state using in-memory runtime configuration
    if (config_) {
        nlohmann::json telemetry_info = {
            {"enabled", config_->telemetry_enabled()}
        };
        if (config_->telemetry_enabled()) {
            std::vector<std::string> captures;
            if (!config_->telemetry_hide_inputs()) {
                captures.push_back("inputs");
            }
            if (!config_->telemetry_hide_outputs()) {
                captures.push_back("outputs");
            }
            if (!config_->telemetry_hide_thinking()) {
                captures.push_back("thinking");
            }
            telemetry_info["captures"] = captures;
        }
        response["telemetry"] = telemetry_info;
    }

    response["high_security"] = !admin_api_key_.empty();

    // Add model loaded information like Python implementation
    std::string loaded_model = router_->get_loaded_model();

    response["model_loaded"] = loaded_model.empty() ? nlohmann::json(nullptr) : nlohmann::json(loaded_model);

    // Multi-model support: Add all loaded models
    response["all_models_loaded"] = router_->get_all_loaded_models();

    // Add max model limits
    response["max_models"] = router_->get_max_model_limits();

    // Add pinned model counts
    response["pinned_models"] = router_->get_pinned_model_counts();
    response["pinned_helper_models"] = router_->get_pinned_helper_counts();

    // Add WebSocket server port for realtime API and log streaming
    if (websocket_server_ && websocket_server_->is_running()) {
        response["websocket_port"] = websocket_server_->get_port();
    }

    // Add update check status
    response["update_check_done"] = update_check_done_.load();

    res.set_content(response.dump(), "application/json");
}

void Server::handle_live(const httplib::Request& req, httplib::Response& res) {
    // For HEAD requests, just return 200 OK
    if (req.method == "HEAD") {
        res.status = 200;
        return;
    }

    // liveness response
    static const char* kLiveResponse = R"({"status":"ok"})";

    res.set_content(kLiveResponse, "application/json");
    res.status = 200;
}

void Server::handle_model_update_check(const httplib::Request& req, httplib::Response& res) {
    (void)req;

    // A manual check is intentionally independent from
    // auto_check_model_updates, but full offline mode remains authoritative.
    if (config_->offline()) {
        res.status = 409;
        res.set_content(
            nlohmann::json{{"error", "Cannot check model updates while offline=true"}}.dump(),
            "application/json");
        return;
    }

    try {
        auto check_res = model_manager_->check_for_model_updates();

        nlohmann::json response = {
            {"status", check_res.failed_models.empty() ? "success" : "failed"},
            {"updates_available", check_res.updated_models.size()},
            {"models", check_res.updated_models},
            {"failed_models", check_res.failed_models}
        };

        res.set_content(response.dump(), "application/json");
    } catch (const std::exception& e) {
        LOG(WARNING, "Server") << "Manual model update check failed: " << e.what() << std::endl;
        res.status = 500;
        res.set_content(
            nlohmann::json{{"error", std::string("Model update check failed: ") + e.what()}}.dump(),
            "application/json");
    }
}

void Server::handle_models_sync(const httplib::Request& req, httplib::Response& res) {
    if (config_->offline()) {
        res.status = 409;
        res.set_content(
            nlohmann::json{{"error", "Cannot sync model updates while offline=true"}}.dump(),
            "application/json");
        return;
    }

    std::vector<std::string> target_models;
    bool dry_run = false;
    bool async_dispatch = true;
    bool async_specified = false;
    bool attach_if_running = false;

    if (!req.body.empty()) {
        try {
            auto body_json = nlohmann::json::parse(req.body);
            if (!body_json.is_object()) {
                res.status = 400;
                res.set_content(
                    nlohmann::json{{"error", "Request body must be a JSON object"}}.dump(),
                    "application/json");
                return;
            }
            if (body_json.contains("models")) {
                if (!body_json["models"].is_array()) {
                    res.status = 400;
                    res.set_content(
                        nlohmann::json{{"error", "'models' must be an array of strings"}}.dump(),
                        "application/json");
                    return;
                }
                for (const auto& item : body_json["models"]) {
                    if (!item.is_string()) {
                        res.status = 400;
                        res.set_content(
                            nlohmann::json{{"error", "Each item in 'models' must be a string"}}.dump(),
                            "application/json");
                        return;
                    }
                    target_models.push_back(item.get<std::string>());
                }
            }
            if (body_json.contains("dry_run")) {
                if (!body_json["dry_run"].is_boolean()) {
                    res.status = 400;
                    res.set_content(
                        nlohmann::json{{"error", "'dry_run' must be a boolean"}}.dump(),
                        "application/json");
                    return;
                }
                dry_run = body_json["dry_run"].get<bool>();
            }
            if (body_json.contains("async")) {
                if (!body_json["async"].is_boolean()) {
                    res.status = 400;
                    res.set_content(
                        nlohmann::json{{"error", "'async' must be a boolean"}}.dump(),
                        "application/json");
                    return;
                }
                async_dispatch = body_json["async"].get<bool>();
                async_specified = true;
            }
            if (body_json.contains("attach_if_running")) {
                if (!body_json["attach_if_running"].is_boolean()) {
                    res.status = 400;
                    res.set_content(
                        nlohmann::json{{"error", "'attach_if_running' must be a boolean"}}.dump(),
                        "application/json");
                    return;
                }
                attach_if_running = body_json["attach_if_running"].get<bool>();
            }
        } catch (const std::exception& e) {
            res.status = 400;
            res.set_content(
                nlohmann::json{{"error", std::string("Invalid JSON body: ") + e.what()}}.dump(),
                "application/json");
            return;
        }
    }

    if (dry_run) {
        try {
            nlohmann::json sync_result = model_manager_->sync_models(target_models, true);
            res.set_content(sync_result.dump(), "application/json");
            res.status = 200;
        } catch (const std::exception& e) {
            LOG(WARNING, "Server") << "Model sync dry-run failed: " << e.what() << std::endl;
            res.status = 500;
            res.set_content(
                nlohmann::json{{"error", std::string("Model sync dry-run failed: ") + e.what()}}.dump(),
                "application/json");
        }
        return;
    }

    if (!async_specified || async_dispatch) {
        auto enqueue_res = model_manager_->enqueue_sync(target_models, attach_if_running);
        if (enqueue_res.already_running) {
            nlohmann::json status = model_manager_->get_sync_status(enqueue_res.sync_id);
            res.set_content(status.dump(), "application/json");
            res.status = 200;
            return;
        }

        std::lock_guard<std::mutex> lock(background_sync_mutex_);

        for (auto it = background_sync_threads_.begin(); it != background_sync_threads_.end(); ) {
            if (it->finished->load() && it->thread.joinable()) {
                it->thread.join();
                it = background_sync_threads_.erase(it);
            } else {
                ++it;
            }
        }

        auto finished_flag = std::make_shared<std::atomic<bool>>(false);
        std::thread worker([this, finished_flag]() {
            try {
                LOG(INFO, "Server") << "Background model sync thread started" << std::endl;
                model_manager_->execute_sync();
                LOG(INFO, "Server") << "Background model sync thread completed successfully" << std::endl;
            } catch (const std::exception& e) {
                LOG(WARNING, "Server") << "Background model sync failed: " << e.what() << std::endl;
            }
            finished_flag->store(true);
        });

        background_sync_threads_.push_back({std::move(worker), finished_flag});

        nlohmann::json dispatched_result = model_manager_->get_sync_status(enqueue_res.sync_id);
        dispatched_result["message"] = "Model synchronization dispatched in background";
        dispatched_result["async"] = true;
        dispatched_result["already_in_progress"] = false;
        res.set_content(dispatched_result.dump(), "application/json");
        res.status = 202;
        return;
    }


    try {
        nlohmann::json sync_result = model_manager_->sync_models(target_models, false, attach_if_running);
        res.set_content(sync_result.dump(), "application/json");
        res.status = 200;
    } catch (const std::exception& e) {
        LOG(WARNING, "Server") << "Model sync failed: " << e.what() << std::endl;
        res.status = 500;
        res.set_content(
            nlohmann::json{{"error", std::string("Model sync failed: ") + e.what()}}.dump(),
            "application/json");
    }
}

void Server::handle_models_sync_status(const httplib::Request& req, httplib::Response& res) {
    try {
        uint64_t sync_id = 0;
        if (req.has_param("sync_id")) {
            try {
                sync_id = std::stoull(req.get_param_value("sync_id"));
            } catch (...) {
                res.status = 400;
                res.set_content(
                    nlohmann::json{{"error", "Invalid sync_id parameter"}}.dump(),
                    "application/json");
                return;
            }
        }
        nlohmann::json status = model_manager_->get_sync_status(sync_id);
        res.set_content(status.dump(), "application/json");
        res.status = 200;
    } catch (const std::exception& e) {
        LOG(WARNING, "Server") << "Failed to get sync status: " << e.what() << std::endl;
        res.status = 500;
        res.set_content(
            nlohmann::json{{"error", std::string("Failed to get sync status: ") + e.what()}}.dump(),
            "application/json");
    }
}

void Server::handle_models(const httplib::Request& req, httplib::Response& res) {
    // For HEAD requests, just return 200 OK without processing
    if (req.method == "HEAD") {
        res.status = 200;
        return;
    }

    // Check if we should show all models (for CLI list command) or only downloaded (OpenAI API behavior)
    bool show_all = req.has_param("show_all") && req.get_param_value("show_all") == "true";

    // OPTIMIZATION: For OpenAI API mode, use get_downloaded_models() which filters first
    // Only use get_supported_models() when we need to show ALL models
    std::map<std::string, ModelInfo> models;
    if (show_all) {
        models = model_manager_->get_supported_models();
    } else {
        models = model_manager_->get_downloaded_models();
    }

    nlohmann::json response;
    response["data"] = nlohmann::json::array();
    response["object"] = "list";

    for (const auto& [model_id, model_info] : models) {
        response["data"].push_back(model_info_to_json(model_id, model_info));
    }

    if (alias_manager_) {
        auto all_aliases = alias_manager_->get_all_aliases();
        for (const auto& [alias_id, target_name] : all_aliases) {
            if (models.count(alias_id) > 0) {
                continue;
            }
            std::string ultimate_target = target_name;
            if (auto resolved = alias_manager_->resolve_alias(alias_id)) {
                ultimate_target = *resolved;
            }
            std::string canonical_target = model_manager_->resolve_model_name(ultimate_target);
            if (models.count(canonical_target) > 0) {
                response["data"].push_back(model_info_to_json(alias_id, models.at(canonical_target)));
            } else if (models.count(ultimate_target) > 0) {
                response["data"].push_back(model_info_to_json(alias_id, models.at(ultimate_target)));
            }
        }
    }

    res.set_content(response.dump(), "application/json");
}

void Server::validate_model_registration_name(const std::string& model_name,
                                             bool require_user_namespace) {
    if (lemon::is_reserved_registration_name(model_name)) {
        throw std::invalid_argument(
            "Model names with 'extra.' / 'builtin.' prefixes are reserved, "
            "including as bare-name parts of a 'user.' alias. Received: " + model_name);
    }

    if (require_user_namespace &&
        (model_name.size() <= 5 || model_name.rfind("user.", 0) != 0)) {
        throw std::invalid_argument(
            "Registered model definitions must use a non-empty `user.*` name, "
            "for example `user.Phi-4-Mini-GGUF`. Received: " + model_name);
    }
}

void Server::normalize_model_registration_source(nlohmann::json& request_json,
                                                 bool local_import) {
    if (!request_json.contains("source") && !request_json.contains("registry_source")) {
        return;
    }

    std::optional<std::string> normalized_registry;
    if (request_json.contains("registry_source")) {
        if (!request_json["registry_source"].is_string()) {
            throw std::invalid_argument("`registry_source` must be a string when provided");
        }
        normalized_registry = remote_registry_source_name(
            parse_remote_registry_source(
                request_json["registry_source"].get<std::string>()));
    }

    if (request_json.contains("source")) {
        if (!request_json["source"].is_string()) {
            throw std::invalid_argument("`source` must be a string when provided");
        }
        const std::string public_source = request_json["source"].get<std::string>();
        if (is_remote_registry_source(public_source)) {
            const std::string normalized_public = remote_registry_source_name(
                parse_remote_registry_source(public_source));
            if (normalized_registry && *normalized_registry != normalized_public) {
                throw std::invalid_argument(
                    "`source` and `registry_source` must identify the same registry");
            }
            normalized_registry = normalized_public;
            request_json["source"] = normalized_public;
        } else if (!local_import && public_source != "local_upload" &&
                   public_source != "local_path" &&
                   public_source != "extra_models_dir") {
            throw std::invalid_argument(
                "Unsupported model source '" + public_source +
                "' (expected 'huggingface', 'modelscope', or a local source)");
        }
    }

    if (normalized_registry) {
        if (!request_json.contains("source") ||
            is_remote_registry_source(request_json["source"].get<std::string>())) {
            request_json["source"] = *normalized_registry;
        }
        request_json["registry_source"] = *normalized_registry;
    }
}

void Server::validate_and_canonicalize_collection_registration(
        const std::string& model_name,
        nlohmann::json& request_json,
        bool allow_embedded_models) {
    const std::string recipe = request_json.value("recipe", std::string());
    if (!is_model_collection_recipe(recipe)) {
        return;
    }

    if (request_json.contains("models") && !allow_embedded_models) {
        throw std::invalid_argument(
            "`models` embeds additional model definitions and is not accepted by "
            "the single-model registration endpoint; register components first");
    }

    if (request_json.contains("components")) {
        if (!request_json["components"].is_array()) {
            throw std::invalid_argument("`components` must be an array when provided");
        }
        for (const auto& component : request_json["components"]) {
            if (!component.is_string()) {
                throw std::invalid_argument(
                    "Every collection component must be a string model name");
            }
        }
    }

    if (auto err = model_manager_->validate_collection_request(model_name, request_json)) {
        throw std::invalid_argument(*err);
    }

    // Single-definition collections reference models already present in the
    // registry. Exported bundles keep their raw component names because their
    // embedded definitions are resolved by the collection import/download path.
    if (request_json.contains("components") && request_json["components"].is_array() &&
        (!request_json.contains("models") || !request_json["models"].is_array())) {
        for (auto& component : request_json["components"]) {
            component = model_manager_->resolve_model_name(component.get<std::string>());
        }
    }
}

std::string Server::register_model_definition_internal(
        const std::string& model_name,
        nlohmann::json& request_json,
        bool require_definition,
        bool allow_embedded_models,
        bool local_import) {
    if (!request_json.is_object()) {
        throw std::invalid_argument("Request body must be a JSON object");
    }

    if (request_json.contains("recipe") && !request_json["recipe"].is_string()) {
        throw std::invalid_argument("`recipe` must be a string when provided");
    }
    const std::string recipe = request_json.value("recipe", std::string());
    if (require_definition && recipe.empty()) {
        throw std::invalid_argument("A non-empty string `recipe` is required");
    }

    if (request_json.contains("checkpoint") && !request_json["checkpoint"].is_string()) {
        throw std::invalid_argument("`checkpoint` must be a string when provided");
    }
    if (request_json.contains("checkpoints")) {
        const auto& checkpoints = request_json["checkpoints"];
        if (!checkpoints.is_object() || !checkpoints.contains("main")) {
            throw std::invalid_argument(
                "If present, `checkpoints` must be an object containing `main`");
        }
        for (const auto& [role, checkpoint] : checkpoints.items()) {
            if (!checkpoint.is_string()) {
                throw std::invalid_argument(
                    "Every `checkpoints` value must be a string; invalid role: " + role);
            }
        }
    }

    const bool has_definition =
        require_definition || !recipe.empty() || request_json.contains("checkpoint") ||
        request_json.contains("checkpoints") || request_json.contains("components") ||
        request_json.contains("models");
    validate_model_registration_name(model_name, has_definition);

    if (request_json.contains("models") && !allow_embedded_models) {
        throw std::invalid_argument(
            "`models` embeds additional model definitions and is not accepted by "
            "the single-model registration endpoint; register components first");
    }

    normalize_model_registration_source(request_json, local_import);
    validate_and_canonicalize_collection_registration(
        model_name, request_json, allow_embedded_models);


    if (!has_definition && !local_import) {
        return model_name;
    }

    if (local_import) {
        std::string hf_cache = model_manager_->get_hf_cache_dir();
        std::string model_name_clean = model_name.substr(5);
        std::replace(model_name_clean.begin(), model_name_clean.end(), '/', '-');
        std::string dest_path = hf_cache + "/models--" + model_name_clean;

        LOG(INFO, "Server") << "Local import mode - resolving files in: "
                            << dest_path << std::endl;
        resolve_and_register_local_model(dest_path, model_name, request_json, hf_cache);
    } else {
        model_manager_->register_model(
            model_name,
            request_json,
            /*allow_missing_checkpoint=*/require_definition,
            /*replace_existing=*/require_definition);
    }

    return model_manager_->get_public_model_name(model_name);
}

void Server::handle_model_register(const httplib::Request& req, httplib::Response& res) {
    auto bad_request = [&res](const std::string& message) {
        res.status = 400;
        res.set_content(nlohmann::json{{"error", message}}.dump(), "application/json");
    };

    nlohmann::json request_json;
    if (!parse_required_json_body(req, res, request_json)) return;

    try {
        if (!request_json.is_object()) {
            bad_request("Request body must be a JSON object");
            return;
        }
        if (!request_json.contains("model_name") || !request_json["model_name"].is_string()) {
            bad_request("A string `model_name` is required");
            return;
        }

        const std::string model_name = request_json["model_name"].get<std::string>();
        const std::string public_name = register_model_definition_internal(
            model_name,
            request_json,
            /*require_definition=*/true,
            /*allow_embedded_models=*/false,
            /*local_import=*/false);

        nlohmann::json response = {
            {"status", "success"},
            {"model_name", public_name},
            {"canonical_model_name", model_name},
        };

        const auto models = model_manager_->get_supported_models();
        auto model_it = models.find(public_name);
        if (model_it != models.end()) {
            response["model"] = model_info_to_json(public_name, model_it->second);
        }

        res.set_content(response.dump(), "application/json");
    } catch (const std::invalid_argument& e) {
        bad_request(e.what());
    } catch (const nlohmann::json::exception& e) {
        bad_request(e.what());
    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "ERROR in handle_model_register: " << e.what() << std::endl;
        res.status = 500;
        res.set_content(nlohmann::json{{"error", e.what()}}.dump(), "application/json");
    }
}

int64_t Server::resolve_context_length(const std::string& model_id, const ModelInfo& info) const {
    // ctx_size stores -1 for "size this automatically", so only a positive
    // value answers; anything else falls through to the next source.
    auto ctx_size_of = [](const RecipeOptions& options) -> int64_t {
        const nlohmann::json ctx_json = options.get_option("ctx_size");
        return ctx_json.is_number() ? ctx_json.get<int64_t>() : 0;
    };

    if (router_) {
        const int64_t loaded_ctx =
            ctx_size_of(router_->get_model_recipe_options(resolve_alias_target(model_id)));
        if (loaded_ctx > 0) {
            return loaded_ctx;
        }

        const RecipeOptions no_request_options(info.recipe, nlohmann::json::object());
        const int64_t configured_ctx =
            ctx_size_of(router_->resolve_effective_options(info, no_request_options));
        if (configured_ctx > 0) {
            return configured_ctx;
        }
    }

    return info.max_context_window > 0 ? info.max_context_window : 0;
}

// Maximum collection-component nesting depth embedded in "models" arrays.
// Collection components are normally leaf models, but nothing prevents
// registering a collection as a component of another collection — including
// cyclically — so the embedding recursion must be bounded.
static constexpr int kMaxCollectionEmbedDepth = 3;

nlohmann::json Server::model_info_to_json(const std::string& model_id, const ModelInfo& info,
                                          int depth) {
    std::vector<std::string> public_components;
    public_components.reserve(info.components.size());
    for (const auto& component : info.components) {
        public_components.push_back(model_manager_->get_public_model_name(component));
    }
    nlohmann::json model_json = {
        {"id", model_id},
        {"object", "model"},
        {"created", 1234567890},
        {"owned_by", "lemonade"},
        {"checkpoint", info.checkpoint()},
        {"checkpoints", info.checkpoints},
        {"recipe", info.recipe},
        {"downloaded", info.downloaded},
        {"update_available", info.update_available},
        {"suggested", info.suggested},
        {"source", info.source.empty() ? info.registry_source : info.source},
        {"registry_source", info.registry_source},
        {"labels", info.labels},
        {"components", public_components},
        {"recipe_options", info.recipe_options.to_json()},
    };

    // Surface the cloud provider on cloud entries so the Model Manager can
    // bucket each provider into its own sub-heading. Omitted on local models
    // so the field doesn't pollute every entry — and skipped for the Ollama
    // serialization path (handle_ollama_show / tags) which builds its own
    // payload.
    if (!info.cloud_provider.empty()) {
        model_json["cloud_provider"] = info.cloud_provider;
    }

    // Add size if available
    if (info.size > 0.0) {
        model_json["size"] = info.size;
    }

    if (info.max_context_window > 0) {
        model_json["max_context_window"] = info.max_context_window;
    }

    // OpenAI-compatible clients use context_length to set token limits.
    const int64_t context_length = resolve_context_length(model_id, info);
    if (context_length > 0) {
        model_json["context_length"] = context_length;
    }

    // Per-million-token pricing in USD, when the provider reported it (cloud
    // models from OpenRouter/Together). Display only.
    if (info.cost_input_per_million >= 0) {
        model_json["cost_input_per_million"] = info.cost_input_per_million;
    }
    if (info.cost_output_per_million >= 0) {
        model_json["cost_output_per_million"] = info.cost_output_per_million;
    }

    // Per-collection system prompt override (collection.omni only). Omitted on
    // models that don't carry one so the field doesn't pollute every entry.
    if (!info.system_prompt.empty()) {
        model_json["system_prompt"] = info.system_prompt;
    }

<<<<<<<
    // User-defined metadata must round-trip through /models. GUI3 uses these
    // fields to reconstruct and edit Omni collections after an application or
    // server restart instead of relying on renderer-local storage.
    for (const auto& label : info.labels) {
        if (label == "custom") {
            model_json["custom"] = true;
            break;
        }
    }
    for (const char* key : {"display_name", "component_roles", "custom_tools"}) {
        auto it = info.extras.find(key);
        if (it != info.extras.end() && !it->second.is_null()) {
            model_json[key] = it->second;
        }
=======
    auto audio_defaults = info.extras.find("audio_defaults");
    if (audio_defaults != info.extras.end() &&
        audio_defaults->second.is_object()) {
        model_json["audio_defaults"] = audio_defaults->second;
>>>>>>>
    }

    // Add image_defaults if present (for sd-cpp models)
    if (info.image_defaults.has_defaults) {
        json img_def = {
            {"steps", info.image_defaults.steps},
            {"cfg_scale", info.image_defaults.cfg_scale},
            {"width", info.image_defaults.width},
            {"height", info.image_defaults.height}
        };
        if (!info.image_defaults.sampling_method.empty())
            img_def["sampling_method"] = info.image_defaults.sampling_method;
        if (info.image_defaults.flow_shift > 0.0f)
            img_def["flow_shift"] = info.image_defaults.flow_shift;
        model_json["image_defaults"] = img_def;
    }

    if (is_router_collection_recipe(info.recipe)) {
        // The parser requires a root "version"; surface it alongside "routing"
        // so an exported router collection can be re-imported through /pull.
        auto version_it = info.extras.find("version");
        if (version_it != info.extras.end()) {
            model_json["version"] = version_it->second;
        }
        auto routing_it = info.extras.find("routing");
        if (routing_it != info.extras.end() && routing_it->second.is_object()) {
            model_json["routing"] = routing_it->second;
        }
    }

    // Collections additionally embed each component's full model object,
    // in component order, under "models". Embedding is bounded by
    // kMaxCollectionEmbedDepth so nested (or cyclic) collection registrations
    // cannot recurse unboundedly.
    if (is_model_collection_recipe(info.recipe) && depth < kMaxCollectionEmbedDepth) {
        nlohmann::json component_models = nlohmann::json::array();
        for (const auto& component : info.components) {
            if (!model_manager_->model_exists(component)) {
                continue;
            }
            auto comp_info = model_manager_->get_model_info(component);
            component_models.push_back(model_info_to_json(
                model_manager_->get_public_model_name(component), comp_info, depth + 1));
        }
        model_json["models"] = component_models;
    }

    return model_json;
}

void Server::handle_model_by_id(const httplib::Request& req, httplib::Response& res) {
    std::string model_id = utils::normalize_model_name(req.matches[1]);

    if (model_manager_->model_exists(model_id)) {
        auto info = model_manager_->get_model_info(model_id);
        std::string canonical_cache_key = model_manager_->resolve_model_name(model_id);
        std::string wire_id = model_manager_->get_public_model_name(canonical_cache_key);
        if (alias_manager_ && alias_manager_->has_alias(model_id)) {
            wire_id = model_id;
        }
        res.set_content(model_info_to_json(wire_id, info).dump(), "application/json");
    } else if (alias_manager_ && alias_manager_->has_alias(model_id)) {
        auto resolved_target = alias_manager_->resolve_alias(model_id);
        if (resolved_target) {
            std::string canonical_target = model_manager_->resolve_model_name(*resolved_target);
            if (model_manager_->model_exists(canonical_target)) {
                auto info = model_manager_->get_model_info(canonical_target);
                res.set_content(model_info_to_json(model_id, info).dump(), "application/json");
                return;
            } else if (model_manager_->model_exists(*resolved_target)) {
                auto info = model_manager_->get_model_info(*resolved_target);
                res.set_content(model_info_to_json(model_id, info).dump(), "application/json");
                return;
            }
        }
        res.status = 404;
        auto error_response = create_model_error(model_id, "Model not found");
        res.set_content(error_response.dump(), "application/json");
    } else {
        res.status = 404;
        auto error_response = create_model_error(model_id, "Model not found");
        res.set_content(error_response.dump(), "application/json");
    }
}

void Server::handle_model_files(const httplib::Request& req, httplib::Response& res) {
    std::string model_id = req.matches[1];
    const bool include_paths = req.has_param("include_paths") &&
        req.get_param_value("include_paths") == "true";

    try {
        if (!model_manager_->model_exists(model_id)) {
            res.status = 404;
            auto error_response = create_model_error(model_id, "Model not found");
            res.set_content(error_response.dump(), "application/json");
            return;
        }

        std::string canonical_cache_key = model_manager_->resolve_model_name(model_id);
        std::string wire_id = model_manager_->get_public_model_name(canonical_cache_key);
        auto files = model_manager_->list_model_files(model_id);

        nlohmann::json response;
        response["model_id"] = wire_id;
        response["files"] = nlohmann::json::array();

        for (const auto& file : files) {
            nlohmann::json file_json = {
                {"name", file.name},
                {"role", file.role},
                {"size_bytes", file.size_bytes},
                {"exists", file.exists}
            };

            if (include_paths) {
                file_json["path"] = file.path;
            }

            response["files"].push_back(std::move(file_json));
        }

        res.set_content(response.dump(), "application/json");
    } catch (const std::exception&) {
        res.status = 404;
        auto error_response = create_model_error(model_id, "Model not found");
        res.set_content(error_response.dump(), "application/json");
    }
}

// `pinned` is live-process state: Router::load_model keeps a running server's
// own pin rather than the resolved one, so a value saved here would not reach a
// model that is already up. /v1/load and /internal/pin own it, and this endpoint
// neither reports it in effective/defaults nor accepts it.
static bool is_live_process_option(const std::string& key) {
    return key == "pinned";
}

// Fill in every option the recipe accepts, resolving unset keys through the
// default chain, so a client can render a complete form from one response.
static nlohmann::json resolve_all_recipe_options(const RecipeOptions& options) {
    nlohmann::json resolved = options.to_resolved_json();
    resolved.erase("pinned");
    return resolved;
}

static bool option_type_matches(const nlohmann::json& expected, const nlohmann::json& value) {
    if (expected.is_number()) return value.is_number();
    if (expected.is_string()) return value.is_string();
    // Booleans, and the tri-state options whose default is null to mean "follow
    // the global setting", both only ever hold a boolean once set.
    return value.is_boolean();
}

// Reject values that pass the type check but would break the load they are
// saved for. Returns an error message, or empty when the value is usable.
static std::string validate_option_value(const std::string& key,
                                         const nlohmann::json& expected,
                                         const nlohmann::json& value) {
    if (key == "ctx_size") {
        // A literal above INT64_MAX parses as an unsigned and wraps on the way
        // to int64_t, so 2^64-1 would arrive as -1 and read as "size it
        // automatically". Rule it out before the conversion.
        const bool fits_int64 = value.is_number_integer() &&
            (!value.is_number_unsigned() ||
             value.get<uint64_t>() <= static_cast<uint64_t>(INT64_MAX));
        // 0 is not a shorthand for anything here: only -1 auto-resolves, so a 0
        // reaches the backend verbatim and FastFlowLM and vLLM both reject it.
        if (!fits_int64 || value.get<int64_t>() < -1 || value == 0) {
            return "'ctx_size' must be a positive whole number, "
                   "or -1 to size it automatically";
        }
        return "";
    }

    // A fractional value for a whole-number option is silently ignored by the
    // consumers that read it, so refuse it.
    if (expected.is_number_integer() && !value.is_number_integer()) {
        return "'" + key + "' must be a whole number";
    }
    return "";
}

void Server::respond_with_model_options(
    const httplib::Request& req, httplib::Response& res,
    const std::function<bool(const std::string&, ModelInfo&, httplib::Response&)>& mutation) {
    const std::string model_id = utils::normalize_model_name(req.matches[1]);
    std::string model_key = model_id;

    if (!model_manager_->model_exists(model_key)) {
        std::optional<std::string> target;
        if (alias_manager_ && alias_manager_->has_alias(model_id)) {
            target = alias_manager_->resolve_alias(model_id);
        }
        std::string canonical_target = target ? model_manager_->resolve_model_name(*target) : "";
        if (target && model_manager_->model_exists(canonical_target)) {
            model_key = canonical_target;
        } else if (target && model_manager_->model_exists(*target)) {
            model_key = *target;
        } else {
            res.status = 404;
            res.set_content(create_model_error(model_id, "Model not found").dump(), "application/json");
            return;
        }
    }

    // Work in canonical names from here on: a user model's requested id is its
    // bare public name, which not every downstream lookup resolves for itself.
    model_key = model_manager_->resolve_model_name(model_key);

    try {
        ModelInfo info = model_manager_->get_model_info(model_key);
        if (mutation && !mutation(model_key, info, res)) return;

        const RecipeOptions no_request_options(info.recipe, nlohmann::json::object());
        RecipeOptions effective = router_->resolve_effective_options(info, no_request_options);

        ModelInfo without_saved = info;
        without_saved.recipe_options = model_manager_->get_model_default_options(info);
        RecipeOptions defaults = router_->resolve_effective_options(without_saved, no_request_options);

        // `effective` and `defaults` double as replayable /v1/load bodies.
        nlohmann::json effective_json = resolve_all_recipe_options(effective);
        nlohmann::json defaults_json = resolve_all_recipe_options(defaults);
        effective_json["model_name"] = model_id;
        defaults_json["model_name"] = model_id;

        const int64_t auto_ctx = resolve_auto_ctx_size(effective, info);
        const nlohmann::json effective_ctx = effective.get_option("ctx_size");
        const int64_t resolved_ctx = auto_ctx != -2 ? auto_ctx
            : (effective_ctx.is_number() ? effective_ctx.get<int64_t>() : -1);

        nlohmann::json response = {
            {"model_name", model_id},
            {"recipe", info.recipe},
            {"saved", model_manager_->get_saved_model_options(model_key)},
            {"effective", effective_json},
            {"defaults", defaults_json},
            {"resolved_ctx_size", resolved_ctx}
        };
        res.set_content(response.dump(), "application/json");
    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "Failed to handle options for '" << model_id
                             << "': " << e.what() << std::endl;
        if (!model_manager_->model_exists(model_key)) {
            // The model went away between the existence check and the read.
            res.status = 404;
            res.set_content(create_model_error(model_key, e.what()).dump(), "application/json");
            return;
        }
        // Anything else is a server-side failure, most often the options file
        // being unwritable. Reporting it through create_model_error would call
        // it a load failure, which this endpoint never performs.
        res.status = 500;
        nlohmann::json error = {{"error", {
            {"message", "Failed to read or update options for '" + model_id + "': " + e.what()},
            {"type", "server_error"},
            {"code", "internal_error"}
        }}};
        res.set_content(error.dump(), "application/json");
    }
}

void Server::handle_model_options_get(const httplib::Request& req, httplib::Response& res) {
    respond_with_model_options(req, res, nullptr);
}

void Server::handle_model_options_post(const httplib::Request& req, httplib::Response& res) {
    respond_with_model_options(req, res,
        [this, &req](const std::string& model_key, ModelInfo& info, httplib::Response& r) {
            nlohmann::json body;
            if (!parse_required_json_body(req, r, body)) return false;
            if (!body.is_object()) {
                r.status = 400;
                r.set_content(nlohmann::json{{"error", "Request body must be a JSON object of recipe options"}}
                                  .dump(), "application/json");
                return false;
            }

            bool dry_run = false;
            if (body.contains("dry_run")) {
                if (!body["dry_run"].is_boolean()) {
                    r.status = 400;
                    r.set_content(nlohmann::json{{"error", "'dry_run' must be a boolean"}}
                                      .dump(), "application/json");
                    return false;
                }
                dry_run = body["dry_run"].get<bool>();
                body.erase("dry_run");
            }

            const auto keys = RecipeOptions::keys_for_recipe(info.recipe);
            const std::set<std::string> allowed(keys.begin(), keys.end());
            const RecipeOptions unset(info.recipe, nlohmann::json::object());

            // Validate everything before writing anything, and express each
            // change as set-or-erase so the merge can happen atomically.
            nlohmann::json changes = nlohmann::json::object();
            for (const auto& [key, value] : body.items()) {
                // The URL names the model; tolerate the body's copy so the
                // `effective` object (a /v1/load body) can be posted back here.
                if (key == "model_name") continue;
                if (!allowed.count(key)) {
                    r.status = 400;
                    r.set_content(nlohmann::json{{"error", "Unknown option '" + key +
                                                           "' for recipe '" + info.recipe + "'"}}
                                      .dump(), "application/json");
                    return false;
                }
                if (is_live_process_option(key)) {
                    r.status = 400;
                    r.set_content(nlohmann::json{{"error", "'" + key + "' applies to a running "
                        "model; use /v1/load or /internal/pin"}}.dump(), "application/json");
                    return false;
                }
                // null removes the key. The other values the option system
                // reads as "not set" ("", -1, "auto") still clear every option
                // except ctx_size, where -1 is a value and a malformed size is
                // an error rather than an absence.
                const bool clears = value.is_null() ||
                    (key != "ctx_size" && RecipeOptions::is_default_sentinel(key, value));
                if (clears) {
                    changes[key] = nullptr;
                    continue;
                }
                const nlohmann::json expected = unset.get_option(key);
                if (!option_type_matches(expected, value)) {
                    r.status = 400;
                    r.set_content(nlohmann::json{{"error", "Invalid type for option '" + key + "'"}}
                                      .dump(), "application/json");
                    return false;
                }
                const std::string invalid = validate_option_value(key, expected, value);
                if (!invalid.empty()) {
                    r.status = 400;
                    r.set_content(nlohmann::json{{"error", invalid}}.dump(), "application/json");
                    return false;
                }
                changes[key] = value;
            }

            if (dry_run) {
                info.recipe_options = model_manager_->preview_saved_model_options(info, changes);
                return true;
            }
            if (!changes.empty()) {
                model_manager_->update_saved_model_options(model_key, changes);
            }
            info = model_manager_->get_model_info(model_key);
            return true;
        });
}

void Server::handle_model_options_delete(const httplib::Request& req, httplib::Response& res) {
    respond_with_model_options(req, res,
        [this](const std::string& model_key, ModelInfo& info, httplib::Response&) {
            if (!model_manager_->get_saved_model_options(model_key).empty()) {
                model_manager_->set_saved_model_options(model_key, nlohmann::json::object());
            }
            info = model_manager_->get_model_info(model_key);
            return true;
        });
}

void Server::handle_collection_chat_completions(const nlohmann::json& request_json,
                                                const ModelInfo& collection_info,
                                                httplib::Response& res) {
    // Load the whole collection up front (shared, collection-aware loader) so the
    // model is fully ready — not just its chat component.
    try {
        ensure_collection_loaded(collection_info);
    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "Failed to load collection '" << collection_info.model_name
                             << "': " << e.what() << std::endl;
        auto error_response = create_model_error(collection_info.model_name, e.what());
        std::string error_code = error_response["error"]["code"].get<std::string>();
        res.status = get_http_status_from_error(error_code);
        res.set_content(error_response.dump(), "application/json");
        return;
    }

    const bool is_streaming = request_json.contains("stream") &&
                              request_json["stream"].is_boolean() &&
                              request_json["stream"].get<bool>();

    if (is_streaming) {
        LOG(INFO, "Server") << "POST /api/v1/chat/completions - Collection (streaming): "
                            << collection_info.model_name << std::endl;
        res.set_header("Cache-Control", "no-cache");
        res.set_header("Connection", "keep-alive");
        res.set_header("X-Accel-Buffering", "no");
        res.set_chunked_content_provider(
            "text/event-stream",
            [this, request_json, collection_info](size_t offset, httplib::DataSink& sink) {
                if (offset > 0) return false;  // single pass
                CollectionOrchestrator orchestrator(
                    *router_, *model_manager_,
                    [this](const std::string& m) { auto_load_model_if_needed(m); });
                try {
                    orchestrator.chat_completion_stream(request_json, collection_info, sink);
                } catch (const std::exception& e) {
                    LOG(ERROR, "Server") << "Collection streaming failed: " << e.what() << std::endl;
                }
                return false;
            });
        return;
    }

    LOG(INFO, "Server") << "POST /api/v1/chat/completions - Collection: "
                        << collection_info.model_name << std::endl;
    CollectionOrchestrator orchestrator(
        *router_, *model_manager_,
        [this](const std::string& m) { auto_load_model_if_needed(m); });
    json response = orchestrator.chat_completion(request_json, collection_info);
    if (response.contains("error")) {
        set_error_response(response, res);
        return;
    }
    res.set_content(response.dump(), "application/json");
}

std::optional<RouterDispatchResult> Server::route_collection_request(
    const nlohmann::json& request_json,
    const ModelInfo& collection_info) {
    // The policy is parsed once when the models cache is built (ModelManager),
    // so dispatch just reads it here. A missing policy means the collection
    // failed to parse at cache-build time; return nullopt so the caller leaves
    // the request's model field untouched (fail open).
    if (!collection_info.route_policy) {
        LOG(WARNING, "Server") << "Router collection '" << collection_info.model_name
                               << "' has no parsed routing policy" << std::endl;
        return std::nullopt;
    }

    // The engine owns its policy (and is rebuilt per request because its
    // classifier services are bound to the live Router), so copy the shared,
    // immutable policy into it.
    RoutePolicy policy = *collection_info.route_policy;
    ClassifierServices services = make_router_classifier_services(
        *router_, [this](const std::string& m) {
            auto_load_model_if_needed(m, json::object(),
                                      LoadPurpose::RoutingDependency);
        });
    CostServices cost_services = make_router_cost_services(*router_);
    RoutingPolicyEngine engine(std::move(policy), std::move(services),
                               std::move(cost_services));

    RouteContext ctx = build_route_context(request_json, collection_info.model_name);
    const bool want_trace = request_json.value("route_trace", false);
    Decision decision = engine.route(ctx, want_trace);
    RouterDispatchResult result;
    result.requested_model = collection_info.model_name;
    result.selected_model = decision.route_to;
    result.decision = std::move(decision);
    router_->note_route_decision(utils::conversation_fingerprint(request_json),
                                 result.selected_model);
    return result;
}

void Server::handle_routing_validate(const httplib::Request& req, httplib::Response& res) {
    nlohmann::json request_json;
    if (!parse_required_json_body(req, res, request_json)) return;

    if (!request_json.contains("policy") || !request_json["policy"].is_object()) {
        res.status = 400;
        nlohmann::json error = {{"error", "'policy' must be a JSON object"}};
        res.set_content(error.dump(), "application/json");
        return;
    }
    if (request_json.contains("prompt") && !request_json["prompt"].is_string()) {
        res.status = 400;
        nlohmann::json error = {{"error", "'prompt' must be a string"}};
        res.set_content(error.dump(), "application/json");
        return;
    }
    const std::string prompt = request_json.value("prompt", std::string());

    if (request_json.contains("has_images") && !request_json["has_images"].is_boolean()) {
        res.status = 400;
        nlohmann::json error = {{"error", "'has_images' must be a boolean"}};
        res.set_content(error.dump(), "application/json");
        return;
    }
    const bool has_images = request_json.value("has_images", false);

    if (request_json.contains("has_tools") && !request_json["has_tools"].is_boolean()) {
        res.status = 400;
        nlohmann::json error = {{"error", "'has_tools' must be a boolean"}};
        res.set_content(error.dump(), "application/json");
        return;
    }
    const bool has_tools = request_json.value("has_tools", false);

    std::map<std::string, std::string> metadata;
    if (request_json.contains("metadata")) {
        const nlohmann::json& metadata_json = request_json["metadata"];
        if (!metadata_json.is_object()) {
            res.status = 400;
            nlohmann::json error = {{"error", "'metadata' must be a JSON object of string values"}};
            res.set_content(error.dump(), "application/json");
            return;
        }
        for (auto it = metadata_json.begin(); it != metadata_json.end(); ++it) {
            if (!it.value().is_string()) {
                res.status = 400;
                nlohmann::json error = {{"error",
                    "'metadata' value for key '" + it.key() + "' must be a string"}};
                res.set_content(error.dump(), "application/json");
                return;
            }
            metadata[it.key()] = it.value().get<std::string>();
        }
    }

    try {
        // Ad-hoc validation: the policy isn't attached to a registered model,
        // so accept any candidate/component name as-is rather than requiring
        // it to resolve against the live model registry. This identity
        // resolver is the only relaxation from the normal registration path —
        // require_declared_components stays at its default (true) so the
        // document must still be internally consistent (every candidate,
        // default_model, rule route_to, and classifier model must be listed
        // in collection.components), matching what registration would accept.
        RoutingPolicyParseOptions options;
        options.resolve_component = [](const std::string& name) -> std::optional<std::string> {
            return name;
        };
        nlohmann::json normalized_routing;
        RoutePolicy policy = parse_route_policy_collection(
            request_json["policy"], options, &normalized_routing);

        ClassifierServices services = make_router_classifier_services(
            *router_, [this](const std::string& m) {
                auto_load_model_if_needed(m, json::object(),
                                          LoadPurpose::RoutingDependency);
            });
        RoutingPolicyEngine engine(std::move(policy), std::move(services));

        RouteContext ctx;
        ctx.input = prompt;
        ctx.params.chars = prompt.size();
        // A test prompt is a single turn with no history, so its whole-request
        // total is the prompt itself — the same equality build_route_context
        // gives the history-less `prompt` form. Leaving this at 0 would make
        // every min_total_chars rule silently untestable here.
        ctx.params.total_chars = prompt.size();
        ctx.params.has_images = has_images;
        ctx.params.has_tools = has_tools;
        ctx.metadata = std::move(metadata);

        Decision decision = engine.route(ctx, /*want_trace=*/true);

        // Echo back the policy actually evaluated (`routing.router` desugared
        // into explicit `classifiers`/`rules`, e.g. `__route_0`) so a caller
        // can match `decision.matched_rule` against a rule that exists —
        // the as-authored policy may only have `routing.router` and no
        // `routing.rules` at all.
        nlohmann::json normalized_policy = request_json["policy"];
        normalized_policy["routing"] = std::move(normalized_routing);

        nlohmann::json response = {
            {"decision", route_decision_to_json(decision)},
            {"normalized_policy", std::move(normalized_policy)},
        };
        res.set_content(response.dump(), "application/json");
    } catch (const std::exception& e) {
        res.status = 400;
        nlohmann::json error = {{"error", std::string("Invalid routing policy: ") + e.what()}};
        res.set_content(error.dump(), "application/json");
    }
}

std::optional<RouterDispatchResult> Server::apply_router_collection_dispatch(
    nlohmann::json& request_json) {
    if (!request_json.contains("model") || !request_json["model"].is_string()) {
        return std::nullopt;
    }
    const std::string requested_model = request_json["model"].get<std::string>();
    try {
        if (!model_manager_->model_exists(requested_model)) {
            return std::nullopt;
        }
        ModelInfo info = model_manager_->get_model_info(requested_model);
        if (!is_router_collection_recipe(info.recipe)) {
            return std::nullopt;
        }
        auto dispatch = route_collection_request(request_json, info);
        if (!dispatch) {
            return std::nullopt;
        }
        dispatch->requested_model = requested_model;
        LOG(INFO, "Server") << "Router collection '" << requested_model << "' -> '"
                            << dispatch->selected_model << "'" << std::endl;
        request_json["model"] = dispatch->selected_model;
        request_json.erase("route_trace");
        return dispatch;
    } catch (const RouterResidencyConflictException&) {
        // This is a deterministic hardware-policy conflict, not a routing miss.
        // Let the endpoint serialize it as HTTP 409 instead of silently falling
        // back to trying to load the collection recipe itself.
        throw;
    } catch (const std::exception& e) {
        LOG(WARNING, "Server") << "Router collection dispatch failed for '"
                               << requested_model << "': " << e.what() << std::endl;
    }
    return std::nullopt;
}

std::set<std::string> Server::active_policy_helper_models() {
    std::set<std::string> needed;
    for (const auto& [name, info] : model_manager_->get_supported_models()) {
        (void)name;
        if (!info.route_policy) {
            continue;
        }
        for (const auto& helper : info.route_policy->helper_models) {
            needed.insert(helper);
        }
    }
    return needed;
}

void Server::record_response_telemetry(const nlohmann::json& response,
                                       const nlohmann::json& request_json) {
    if (!response.is_object() ||
        (!response.contains("timings") && !response.contains("usage"))) {
        return;
    }

    StreamingProxy::TelemetryData telemetry = StreamingProxy::extract_telemetry(response);
    std::string model_name = request_json.value("model", "");
    LOG(INFO, "Telemetry") << "Inference completed: model=" << model_name
                           << ", tokens=" << (telemetry.input_tokens + telemetry.output_tokens)
                           << " (in=" << telemetry.input_tokens
                           << ", out=" << telemetry.output_tokens << ")"
                           << ", ttft=" << std::fixed << std::setprecision(2)
                           << telemetry.time_to_first_token << "s"
                           << ", tps=" << telemetry.tokens_per_second << std::endl;

    router_->update_request_telemetry(model_name, telemetry);
}

void Server::handle_chat_completions(const httplib::Request& req, httplib::Response& res) {
    nlohmann::json request_json;
    if (!parse_required_json_body(req, res, request_json)) return;

    try {

        // Normalize client-provided model names (e.g., strip ":latest" suffix)
        // Must be done before any model_manager/router lookups and before forwarding
        normalize_client_model_name(request_json);
        normalize_and_resolve_request_model(request_json);

        // Debug: Check if tools are present
        if (request_json.contains("tools")) {
            LOG(DEBUG, "Server") << "Tools present in request: " << request_json["tools"].size() << " tool(s)" << std::endl;
            LOG(DEBUG, "Server") << "Tools JSON: " << request_json["tools"].dump() << std::endl;
        } else {
            LOG(DEBUG, "Server") << "No tools in request" << std::endl;
        }

        std::optional<RouterDispatchResult> route_dispatch;

        // Omni "collection" models run a server-side tool-calling loop instead of a
        // plain completion. Branch before auto-load/LLM-type checks: the collection
        // recipe has no backend of its own; the orchestrator loads each component.
        if (request_json.contains("model") && request_json["model"].is_string()) {
            const std::string requested_model = request_json["model"].get<std::string>();
            try {
                if (model_manager_->model_exists(requested_model)) {
                    ModelInfo info = model_manager_->get_model_info(requested_model);
                    if (is_omni_collection_recipe(info.recipe)) {
                        handle_collection_chat_completions(request_json, info, res);
                        return;
                    }
                    if (is_router_collection_recipe(info.recipe)) {
                        // The recipe is the trigger (no "auto", no /v1/route): run the
                        // routing engine and rewrite the model to the selected
                        // candidate, then fall through to normal completion handling.
                        route_dispatch = route_collection_request(request_json, info);
                        if (route_dispatch) {
                            route_dispatch->requested_model = requested_model;
                            LOG(INFO, "Server") << "Router collection '" << requested_model
                                                << "' -> '" << route_dispatch->selected_model << "'" << std::endl;
                            request_json["model"] = route_dispatch->selected_model;
                            request_json.erase("route_trace");
                        }
                    }
                }
            } catch (const RouterResidencyConflictException& e) {
                LOG(WARNING, "Server") << "Router residency conflict for '"
                                       << requested_model << "': " << e.what() << std::endl;
                set_router_residency_conflict_response(e, res);
                return;
            } catch (const std::exception& e) {
                LOG(DEBUG, "Server") << "Collection check failed for '" << requested_model
                                     << "': " << e.what() << std::endl;
            }
        }

        std::string requested_model;
        if (request_json.contains("model") && request_json["model"].is_string()) {
            requested_model = request_json["model"].get<std::string>();
        }
        auto span = telemetry::TelemetryTracker::start_span("LLM", "chat.completions", requested_model, request_json);

        // Handle model loading/switching
        if (request_json.contains("model")) {
            try {
                auto_load_model_if_needed(requested_model, extract_auto_load_options(request_json));
                if (span) {
                    span->cancel();
                }
            } catch (const std::exception& e) {
                LOG(ERROR, "Server") << "Failed to load model: " << e.what() << std::endl;
                auto error_response = create_model_error(requested_model, e.what());
                // Set appropriate status code based on error type
                std::string error_code = error_response["error"]["code"].get<std::string>();
                res.status = get_http_status_from_error(error_code);
                res.set_content(error_response.dump(), "application/json");

                if (span) {
                    span->end_with_error(e.what());
                }
                return;
            }
        } else if (!router_->is_model_loaded()) {
            LOG(ERROR, "Server") << "No model loaded and no model specified in request" << std::endl;
            res.status = 400;
            res.set_content("{\"error\": \"No model loaded and no model specified in request\"}", "application/json");
            if (span) {
                span->cancel();
            }
            return;
        }

        // Check if the loaded model supports chat completion (only LLM models do)
        std::string model_to_check = request_json.contains("model") ? request_json["model"].get<std::string>() : "";
        if (router_->get_model_type(model_to_check) != ModelType::LLM) {
            LOG(ERROR, "Server") << "Model does not support chat completion" << std::endl;
            res.status = 400;
            res.set_content(R"({"error": {"message": "This model does not support chat completion. Only LLM models support this endpoint.", "type": "invalid_request_error"}})", "application/json");
            if (span) {
                span->cancel();
            }
            return;
        }

        if (span) {
            span->cancel();
        }

        // Check if streaming is requested
        bool is_streaming = request_json.contains("stream") && request_json["stream"].get<bool>();

        // OpenCode and other OpenAI-compatible clients may send thinking=false
        // instead of Lemonade's enable_thinking=false.
        normalize_thinking_controls(request_json);

        if (is_streaming) {
            try {
                std::string request_body = request_json.dump();
                // Log the HTTP request
                LOG(INFO, "Server") << "POST /api/v1/chat/completions - Streaming" << std::endl;

                set_route_decision_sse_content_provider(
                    res,
                    route_dispatch,
                    request_body,
                    [this](const std::string& body, httplib::DataSink& sink) {
                        router_->chat_completion_stream(body, sink);
                    });
            } catch (const std::exception& e) {
                LOG(ERROR, "Server") << "Streaming failed: " << e.what() << std::endl;
                res.status = 500;
                res.set_content("{\"error\":\"Internal server error during streaming\"}", "application/json");
            }
        } else {
            // Log the HTTP request
            LOG(INFO, "Server") << "POST /api/v1/chat/completions - 200 OK" << std::endl;

            auto response = router_->chat_completion(request_json);

            if (response.contains("error")) {
                LOG(ERROR, "Server") << "Backend returned error response: " << response["error"].dump() << std::endl;
                set_error_response(response, res);
                return;
            }

            attach_route_decision(response, res, route_dispatch);
            // Debug: Check if response contains tool_calls
            if (response.contains("choices") && response["choices"].is_array() && !response["choices"].empty()) {
                auto& first_choice = response["choices"][0];
                if (first_choice.contains("message")) {
                    auto& message = first_choice["message"];
                    if (message.contains("tool_calls")) {
                        LOG(DEBUG, "Server") << "Response contains tool_calls: " << message["tool_calls"].dump() << std::endl;
                    } else {
                        LOG(DEBUG, "Server") << "Response message does NOT contain tool_calls" << std::endl;
                        if (message.contains("content")) {
                            LOG(DEBUG, "Server") << "Message content: " << message["content"].get<std::string>().substr(0, 200) << std::endl;
                        }
                    }
                }
            }

            res.set_content(response.dump(), "application/json");

            record_response_telemetry(response, request_json);


        }

    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "Chat completion failed: " << e.what() << std::endl;

        res.status = 500;
        nlohmann::json error = {{"error", e.what()}};
        res.set_content(error.dump(), "application/json");
    }
}

void Server::handle_completions(const httplib::Request& req, httplib::Response& res) {
    try {
        auto request_json = nlohmann::json::parse(req.body);

        // Normalize client-provided model names (e.g., strip ":latest" suffix)
        // Must be done before any model_manager/router lookups and before forwarding
        normalize_client_model_name(request_json);
        normalize_and_resolve_request_model(request_json);

        // A collection.router model flips this endpoint into engine mode: pick a
        // candidate and rewrite the model before the usual load/forward logic.
        std::optional<RouterDispatchResult> route_dispatch =
            apply_router_collection_dispatch(request_json);

        std::string requested_model;
        if (request_json.contains("model") && request_json["model"].is_string()) {
            requested_model = request_json["model"].get<std::string>();
        }
        auto span = telemetry::TelemetryTracker::start_span("LLM", "completions", requested_model, request_json);

        // Handle model loading/switching (same logic as chat_completions)
        if (request_json.contains("model")) {
            try {
                auto_load_model_if_needed(requested_model, extract_auto_load_options(request_json));
                if (span) {
                    span->cancel();
                }
            } catch (const std::exception& e) {
                LOG(ERROR, "Server") << "Failed to load model: " << e.what() << std::endl;
                auto error_response = create_model_error(requested_model, e.what());
                // Set appropriate status code based on error type
                std::string error_code = error_response["error"]["code"].get<std::string>();
                res.status = get_http_status_from_error(error_code);
                res.set_content(error_response.dump(), "application/json");

                if (span) {
                    span->end_with_error(e.what());
                }
                return;
            }
        } else if (!router_->is_model_loaded()) {
            LOG(ERROR, "Server") << "No model loaded and no model specified in request" << std::endl;
            res.status = 400;
            res.set_content("{\"error\": \"No model loaded and no model specified in request\"}", "application/json");
            if (span) {
                span->cancel();
            }
            return;
        }

        // Check if the loaded model supports completion (only LLM models do)
        std::string model_to_check = request_json.contains("model") ? request_json["model"].get<std::string>() : "";
        if (router_->get_model_type(model_to_check) != ModelType::LLM) {
            LOG(ERROR, "Server") << "Model does not support completion" << std::endl;
            res.status = 400;
            res.set_content(R"({"error": {"message": "This model does not support completion. Only LLM models support this endpoint.", "type": "invalid_request_error"}})", "application/json");
            if (span) {
                span->cancel();
            }
            return;
        }

        if (span) {
            span->cancel();
        }

        // Check if streaming is requested
        bool is_streaming = request_json.contains("stream") && request_json["stream"].get<bool>();

        if (is_streaming) {
            try {
                std::string request_body = request_json.dump();
                // Log the HTTP request
                LOG(INFO, "Server") << "POST /api/v1/completions - Streaming" << std::endl;

                set_route_decision_sse_content_provider(
                    res,
                    route_dispatch,
                    request_body,
                    [this](const std::string& body, httplib::DataSink& sink) {
                        router_->completion_stream(body, sink);
                    });

                LOG(INFO, "Server") << "Streaming completed - 200 OK" << std::endl;
                return;

            } catch (const std::exception& e) {
                LOG(ERROR, "Server") << "Streaming failed: " << e.what() << std::endl;
                res.status = 500;
                res.set_content("{\"error\": \"" + std::string(e.what()) + "\"}", "application/json");
                return;
            }
        } else {
            // Non-streaming
            auto response = router_->completion(request_json);

            // Check if response contains an error
            if (response.contains("error")) {
                LOG(ERROR, "Server") << "Backend returned error response: " << response["error"].dump() << std::endl;
                set_error_response(response, res);
                return;
            }

            // Verify response has required fields
            if (!response.contains("choices")) {
                LOG(ERROR, "Server") << "Response missing 'choices' field. Response: " << response.dump() << std::endl;
                res.status = 500;
                nlohmann::json error = {{"error", "Backend returned invalid response format"}};
                res.set_content(error.dump(), "application/json");
                return;
            }

            attach_route_decision(response, res, route_dispatch);
            res.set_content(response.dump(), "application/json");

            record_response_telemetry(response, request_json);
        }

    } catch (const RouterResidencyConflictException& e) {
        LOG(WARNING, "Server") << "Router residency conflict in completions: "
                               << e.what() << std::endl;
        set_router_residency_conflict_response(e, res);
    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "ERROR in handle_completions: " << e.what() << std::endl;
        res.status = 500;
        nlohmann::json error = {{"error", e.what()}};
        res.set_content(error.dump(), "application/json");
    }
}

void Server::handle_embeddings(const httplib::Request& req, httplib::Response& res) {
    try {
        auto request_json = nlohmann::json::parse(req.body);
        normalize_client_model_name(request_json);
        normalize_and_resolve_request_model(request_json);

        std::string requested_model;
        if (request_json.contains("model") && request_json["model"].is_string()) {
            requested_model = request_json["model"].get<std::string>();
        }
        auto span = telemetry::TelemetryTracker::start_span("EMBEDDING", "embeddings", requested_model, request_json);

        // Handle model loading/switching using helper function
        if (request_json.contains("model")) {
            try {
                auto_load_model_if_needed(requested_model, extract_auto_load_options(request_json));
                if (span) {
                    span->cancel();
                }
            } catch (const std::exception& e) {
                LOG(ERROR, "Server") << "Failed to load model: " << e.what() << std::endl;
                auto error_response = create_model_error(requested_model, e.what());
                std::string error_code = error_response["error"]["code"].get<std::string>();
                res.status = get_http_status_from_error(error_code);
                res.set_content(error_response.dump(), "application/json");

                if (span) {
                    span->end_with_error(e.what());
                }
                return;
            }
        } else if (!router_->is_model_loaded()) {
            LOG(ERROR, "Server") << "No model loaded and no model specified in request" << std::endl;
            res.status = 400;
            res.set_content("{\"error\": \"No model loaded and no model specified in request\"}", "application/json");
            if (span) {
                span->cancel();
            }
            return;
        }

        if (span) {
            span->cancel();
        }

        // Call router's embeddings method
        auto response = router_->embeddings(request_json);
        if (response.contains("error")) {
            set_error_response(response, res);
            return;
        }

        res.set_content(response.dump(), "application/json");

    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "ERROR in handle_embeddings: " << e.what() << std::endl;
        res.status = 500;
        nlohmann::json error = {{"error", e.what()}};
        res.set_content(error.dump(), "application/json");
    }
}

void Server::handle_reranking(const httplib::Request& req, httplib::Response& res) {
    try {
        auto request_json = nlohmann::json::parse(req.body);

        std::string requested_model;
        if (request_json.contains("model") && request_json["model"].is_string()) {
            requested_model = request_json["model"].get<std::string>();
        }
        auto span = telemetry::TelemetryTracker::start_span("RERANKER", "reranking", requested_model, request_json);

        // Handle model loading/switching using helper function
        if (request_json.contains("model")) {
            try {
                auto_load_model_if_needed(requested_model, extract_auto_load_options(request_json));
                if (span) {
                    span->cancel();
                }
            } catch (const std::exception& e) {
                LOG(ERROR, "Server") << "Failed to load model: " << e.what() << std::endl;
                auto error_response = create_model_error(requested_model, e.what());
                std::string error_code = error_response["error"]["code"].get<std::string>();
                res.status = get_http_status_from_error(error_code);
                res.set_content(error_response.dump(), "application/json");

                if (span) {
                    span->end_with_error(e.what());
                }
                return;
            }
        } else if (!router_->is_model_loaded()) {
            LOG(ERROR, "Server") << "No model loaded and no model specified in request" << std::endl;
            res.status = 400;
            res.set_content("{\"error\": \"No model loaded and no model specified in request\"}", "application/json");
            if (span) {
                span->cancel();
            }
            return;
        }

        if (span) {
            span->cancel();
        }

        // Call router's reranking method
        auto response = router_->reranking(request_json);
        if (response.contains("error")) {
            set_error_response(response, res);
            return;
        }

        res.set_content(response.dump(), "application/json");

    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "ERROR in handle_reranking: " << e.what() << std::endl;
        res.status = 500;
        nlohmann::json error = {{"error", e.what()}};
        res.set_content(error.dump(), "application/json");
    }
}

void Server::handle_classify(const httplib::Request& req, httplib::Response& res) {
    try {
        nlohmann::json request_json;
        try {
            request_json = nlohmann::json::parse(req.body);
        } catch (const nlohmann::json::parse_error& e) {
            res.status = 400;
            nlohmann::json error = {{"error", {
                {"message", std::string("Invalid JSON in request body: ") + e.what()},
                {"type", "invalid_request_error"}}}};
            res.set_content(error.dump(), "application/json");
            return;
        }

        normalize_client_model_name(request_json);
        normalize_and_resolve_request_model(request_json);

        // Reject malformed requests before any model gets loaded.
        std::string validation_error;
        bool has_text = request_json.contains("text");
        bool has_input = request_json.contains("input");
        const auto& input_field = has_text ? request_json["text"]
                                 : has_input ? request_json["input"] : request_json;
        if (request_json.contains("model") && !request_json["model"].is_string()) {
            validation_error = "'model' must be a string";
        } else if (!has_text && !has_input) {
            validation_error = "Missing 'input' (or 'text') string in classify request";
        } else if (has_text && !request_json["text"].is_string()) {
            validation_error = "'text' must be a string";
        } else if (has_input && !request_json["input"].is_string()) {
            validation_error = "'input' must be a string";
        } else if (input_field.get<std::string>().find_first_not_of(" \t\r\n") ==
                   std::string::npos) {
            validation_error = "input text must not be empty";
        } else if (request_json.contains("top_k") &&
                   (!request_json["top_k"].is_number_integer() ||
                    request_json["top_k"].get<long long>() < 1 ||
                    request_json["top_k"].get<long long>() > 1000000)) {
            validation_error = "'top_k' must be a positive integer";
        }
        if (!validation_error.empty()) {
            res.status = 400;
            nlohmann::json error = {{"error", {
                {"message", validation_error},
                {"type", "invalid_request_error"}}}};
            res.set_content(error.dump(), "application/json");
            return;
        }

        std::string requested_model;
        if (request_json.contains("model") && request_json["model"].is_string()) {
            requested_model = request_json["model"].get<std::string>();
        }
        auto span = telemetry::TelemetryTracker::start_span("CLASSIFIER", "classify", requested_model, request_json);

        // Handle model loading/switching using helper function
        if (request_json.contains("model")) {
            try {
                auto_load_model_if_needed(requested_model, extract_auto_load_options(request_json));
                if (span) {
                    span->cancel();
                }
            } catch (const std::exception& e) {
                LOG(ERROR, "Server") << "Failed to load model: " << e.what() << std::endl;
                auto error_response = create_model_error(requested_model, e.what());
                std::string error_code = error_response["error"]["code"].get<std::string>();
                res.status = get_http_status_from_error(error_code);
                res.set_content(error_response.dump(), "application/json");

                if (span) {
                    span->end_with_error(e.what());
                }
                return;
            }
        } else {
            // "model" may be omitted only when exactly one classifier is loaded.
            // The router requires a model on every request, so resolve it here
            // and put it in the request rather than letting it fail downstream.
            requested_model = router_->get_sole_loaded_model_of_type(ModelType::CLASSIFICATION);
            if (requested_model.empty()) {
                res.status = 400;
                nlohmann::json error = {{"error", {
                    {"message", "No 'model' specified and no single classification model "
                                "is loaded (load one, or name it in the request)"},
                    {"type", "invalid_request_error"}}}};
                res.set_content(error.dump(), "application/json");
                if (span) {
                    span->cancel();
                }
                return;
            }
            request_json["model"] = requested_model;
        }

        if (span) {
            span->cancel();
        }

        auto response = router_->classify(request_json);
        if (response.contains("error") && response["error"].is_object()) {
            const auto& err = response["error"];
            if (err.contains("status_code") && err["status_code"].is_number_integer()) {
                res.status = err["status_code"].get<int>();
            } else if (err.contains("code") && err["code"].is_string()) {
                res.status = get_http_status_from_error(err["code"].get<std::string>());
            } else if (err.contains("type") && err["type"].is_string()) {
                // LemonException::to_json carries only {message, type}.
                const std::string type = err["type"].get<std::string>();
                if (type == "invalid_request" || type == "invalid_request_error" ||
                    type == "unsupported_operation") {
                    res.status = 400;
                } else if (type == "model_not_loaded") {
                    res.status = 404;
                } else {
                    res.status = 500;
                }
            } else {
                res.status = 500;
            }
            res.set_content(response.dump(), "application/json");
            return;
        }

        // Envelope pins the public API shape; the backend subprocess's raw
        // output is not the contract.
        if (!response.contains("labels") || !response["labels"].is_object()) {
            res.status = 500;
            nlohmann::json error = {{"error", {
                {"message", "Classification backend returned an unexpected response"},
                {"type", "classification_error"}}}};
            res.set_content(error.dump(), "application/json");
            return;
        }
        nlohmann::json enveloped = {
            {"object", "classification"},
            {"model", requested_model.empty() ? router_->get_loaded_model() : requested_model},
            {"labels", response["labels"]},
        };
        res.set_content(enveloped.dump(), "application/json");

    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "ERROR in handle_classify: " << e.what() << std::endl;
        res.status = 500;
        nlohmann::json error = {{"error", e.what()}};
        res.set_content(error.dump(), "application/json");
    }
}

void Server::handle_slots(const httplib::Request& req, httplib::Response& res) {
    try {
        // Slots endpoint doesn't require a model parameter since it queries server state
        // But we still need a model to be loaded to have an active server to query
        if (!router_->is_model_loaded()) {
            LOG(ERROR, "Server") << "No model loaded for slots query" << std::endl;
            res.status = 400;
            res.set_content("{\"error\": \"No model loaded for slots query\"}", "application/json");
            return;
        }

        // Call router's get_slots method
        auto response = router_->get_slots();
        if (response.contains("error")) {
            set_error_response(response, res);
            return;
        }

        res.set_content(response.dump(), "application/json");

    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "ERROR in handle_slots: " << e.what() << std::endl;
        res.status = 500;
        nlohmann::json error = {{"error", e.what()}};
        res.set_content(error.dump(), "application/json");
    }
}

void Server::handle_slots_by_id(const httplib::Request& req, httplib::Response& res) {
    try {
        // Extract slot ID from path parameter
        std::string slot_id_str = req.matches[1];
        int slot_id;
        try {
            slot_id = std::stoi(slot_id_str);
        } catch (const std::exception& e) {
            LOG(ERROR, "Server") << "Invalid slot ID: " << slot_id_str << std::endl;
            res.status = 400;
            res.set_content("{\"error\": \"Invalid slot ID: " + slot_id_str + "\"}", "application/json");
            return;
        }

        // Check for action query parameter
        auto action_param = req.get_param_value("action");
        if (action_param.empty()) {
            LOG(ERROR, "Server") << "Missing action parameter for slots POST endpoint" << std::endl;
            res.status = 400;
            res.set_content("{\"error\": \"POST /api/v1/slots/{id} requires action query parameter (e.g., ?action=erase)\"}", "application/json");
            return;
        }

        // Validate known actions for specific slots (can be extended as needed)
        if (action_param != "erase" && action_param != "save" && action_param != "restore") {
            LOG(ERROR, "Server") << "Unknown action parameter: " << action_param << std::endl;
            res.status = 400;
            res.set_content("{\"error\": \"Unknown action: " + action_param + ". Supported actions: erase, save, restore\"}", "application/json");
            return;
        }

        // Slots actions don't require a model parameter since they operate on server state
        // But we still need a model to be loaded to have an active server to operate on
        if (!router_->is_model_loaded()) {
            LOG(ERROR, "Server") << "No model loaded for slots " << action_param << " operation" << std::endl;
            res.status = 400;
            res.set_content("{\"error\": \"No model loaded for slots " + action_param + " operation\"}", "application/json");
            return;
        }

        // Parse request body as JSON (use empty object if body is empty)
        json request_body;
        if (!req.body.empty()) {
            try {
                request_body = json::parse(req.body);
            } catch (const std::exception& e) {
                LOG(ERROR, "Server") << "Failed to parse request body: " << e.what() << std::endl;
                res.status = 400;
                res.set_content("{\"error\": \"Invalid JSON in request body\"}", "application/json");
                return;
            }
        } else {
            request_body = json::object();
        }

        // Call router's slots_action method with slot ID, action, and request body
        auto response = router_->slots_action(slot_id, action_param, request_body);
        if (response.contains("error")) {
            set_error_response(response, res);
            return;
        }

        res.set_content(response.dump(), "application/json");

    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "ERROR in handle_slots_by_id: " << e.what() << std::endl;
        res.status = 500;
        nlohmann::json error = {{"error", e.what()}};
        res.set_content(error.dump(), "application/json");
    }
}

void Server::handle_tokenize(const httplib::Request& req, httplib::Response& res) {
    try {
        LOG(INFO, "Server") << "POST /api/v1/tokenize" << std::endl;

        // Parse request body as JSON (use empty object if body is empty)
        json request_body;
        if (!req.body.empty()) {
            try {
                request_body = json::parse(req.body);
            } catch (const std::exception& e) {
                LOG(ERROR, "Server") << "Failed to parse request body: " << e.what() << std::endl;
                res.status = 400;
                res.set_content("{\"error\": \"Invalid JSON in request body\"}", "application/json");
                return;
            }
        } else {
            request_body = json::object();
        }

        // Tokenize endpoint requires at least a valid "content" entry in the body
        if (!request_body.contains("content") || !request_body["content"].is_string()) {
            LOG(ERROR, "Server") << "Tokenization failed: 'content' parameter is missing" << std::endl;
            res.status = 400;
            res.set_content("{\"error\": \"'content' parameter is required\"}", "application/json");
            return;
        }

        // Tokenization requires a model to be loaded
        if (!router_->is_model_loaded()) {
            LOG(ERROR, "Server") << "No model loaded for tokenization" << std::endl;
            res.status = 400;
            res.set_content("{\"error\": \"No model loaded for tokenization\"}", "application/json");
            return;
        }

        // Forward request to router
        auto response = router_->tokenize(request_body);
        if (response.contains("error")) {
            set_error_response(response, res);
            return;
        }

        res.set_content(response.dump(), "application/json");

    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "ERROR in handle_tokenize: " << e.what() << std::endl;
        res.status = 500;
        nlohmann::json error = {{"error", e.what()}};
        res.set_content(error.dump(), "application/json");
    }
}

void Server::handle_audio_transcriptions(const httplib::Request& req, httplib::Response& res) {
    try {
        LOG(INFO, "Server") << "POST /api/v1/audio/transcriptions" << std::endl;

        // OpenAI audio API uses multipart form data
        if (!req.is_multipart_form_data()) {
            res.status = 400;
            nlohmann::json error = {{"error", {
                {"message", "Request must be multipart/form-data"},
                {"type", "invalid_request_error"}
            }}};
            res.set_content(error.dump(), "application/json");
            return;
        }

        // Build request JSON for router
        nlohmann::json request_json = nlohmann::json::object();

        // Extract form fields
        if (req.form.has_field("model")) {
            request_json["model"] = req.form.get_field("model");
        }
        normalize_client_model_name(request_json);
        normalize_and_resolve_request_model(request_json);
        if (req.form.has_field("language")) {
            request_json["language"] = req.form.get_field("language");
        }
        if (req.form.has_field("prompt")) {
            request_json["prompt"] = req.form.get_field("prompt");
        }
        if (req.form.has_field("response_format")) {
            request_json["response_format"] = req.form.get_field("response_format");
        }
        if (req.form.has_field("temperature")) {
            request_json["temperature"] = std::stod(req.form.get_field("temperature"));
        }

        // Validate response format early
        std::string response_format =
            request_json.value("response_format", lemon::audio::ResponseFormat::JSON);

        if (!lemon::audio::is_supported_response_format(response_format)) {
            res.status = 400;
            nlohmann::json error = {{"error", {
                {"message", "Unsupported response_format: " + response_format},
                {"type", "invalid_request_error"}
            }}};
            res.set_content(error.dump(), "application/json");
            return;
        }

        // Extract audio file
        const auto& files = req.form.files;
        bool found_audio = false;
        for (const auto& file_pair : files) {
            if (file_pair.first == "file") {
                const auto& file = file_pair.second;
                request_json["file_data"] = file.content;
                request_json["filename"] = file.filename;
                found_audio = true;
                LOG(INFO, "Server") << "Audio file: " << file.filename
                          << " (" << file.content.size() << " bytes)" << std::endl;
                break;
            }
        }

        if (!found_audio) {
            res.status = 400;
            nlohmann::json error = {{"error", {
                {"message", "Missing 'file' field in request"},
                {"type", "invalid_request_error"}
            }}};
            res.set_content(error.dump(), "application/json");
            return;
        }

        // Handle model loading
        if (request_json.contains("model")) {
            std::string requested_model = request_json["model"];
            try {
                auto_load_model_if_needed(requested_model, extract_auto_load_options(request_json));
            } catch (const std::exception& e) {
                LOG(ERROR, "Server") << "Failed to load audio model: " << e.what() << std::endl;
                auto error_response = create_model_error(requested_model, e.what());
                std::string error_code = error_response["error"]["code"].get<std::string>();
                res.status = get_http_status_from_error(error_code);
                res.set_content(error_response.dump(), "application/json");
                return;
            }
        } else {
            res.status = 400;
            nlohmann::json error = {{"error", {
                {"message", "Missing 'model' field in request"},
                {"type", "invalid_request_error"}
            }}};
            res.set_content(error.dump(), "application/json");
            return;
        }

        // Forward to router
        auto response = router_->audio_transcriptions(request_json);

        if (response.contains("error")) {
            set_error_response(response, res, 500);
            return;
        }

        if (lemon::audio::is_plain_text_format(response_format) &&
            response.contains("text") &&
            response["text"].is_string()) {
            res.set_content(response["text"].get<std::string>(), "text/plain");
            return;
        }

        res.set_content(response.dump(), "application/json");

    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "ERROR in handle_audio_transcriptions: " << e.what() << std::endl;
        res.status = 500;
        nlohmann::json error = {{"error", {
            {"message", e.what()},
            {"type", "internal_error"}
        }}};
        res.set_content(error.dump(), "application/json");
    }
}

void Server::handle_audio_speech(const httplib::Request& req, httplib::Response& res) {
    try {
        auto request_json = nlohmann::json::parse(req.body);
        normalize_client_model_name(request_json);
        normalize_and_resolve_request_model(request_json);

        // Handle model loading
        if (request_json.contains("model")) {
            std::string requested_model = request_json["model"];
            if (auto info = router_->try_get_model_info(requested_model);
                info && info->type != ModelType::TTS) {
                res.status = 400;
                res.set_content(nlohmann::json{{"error", {
                    {"message", "Model '" + requested_model +
                        "' is not a text-to-speech model"},
                    {"type", "invalid_request_error"},
                    {"code", "model_not_applicable"}}}}.dump(),
                    "application/json");
                return;
            }
            try {
                auto_load_model_if_needed(requested_model, extract_auto_load_options(request_json));
            } catch (const std::exception& e) {
                LOG(ERROR, "Server") << "Failed to load text-to-speech model: " << e.what() << std::endl;
                auto error_response = create_model_error(requested_model, e.what());
                std::string error_code = error_response["error"]["code"].get<std::string>();
                res.status = get_http_status_from_error(error_code);
                res.set_content(error_response.dump(), "application/json");
                return;
            }
        } else {
            res.status = 400;
            nlohmann::json error = {{"error", {
                {"message", "Missing 'model' field in request"},
                {"type", "invalid_request_error"}
            }}};
            res.set_content(error.dump(), "application/json");
            return;
        }

        if (!request_json.contains("input")) {
            res.status = 400;
            nlohmann::json error = {{"error", {
                {"message", "Missing 'input' field in request"},
                {"type", "invalid_request_error"}
            }}};
            res.set_content(error.dump(), "application/json");
            return;
        }
        if (!request_json["input"].is_string()) {
            res.status = 400;
            nlohmann::json error = {{"error", {
                {"message", "'input' must be a string"},
                {"type", "invalid_request_error"}
            }}};
            res.set_content(error.dump(), "application/json");
            return;
        }

        bool is_streaming = (request_json.contains("stream") && request_json["stream"].get<bool>());

        if (request_json.contains("stream_format")) {
            is_streaming = true;
            if (request_json["stream_format"] != "audio") {
                res.status = 400;
                nlohmann::json error = {{"error", {
                    {"message", "Only 'audio' is supported for stream_format"},
                    {"type", "invalid_request_error"}
                }}};
                res.set_content(error.dump(), "application/json");
                return;
            }
        }

        const std::string speech_model = request_json["model"].get<std::string>();
        // A backend's streaming encoder is often narrower than its buffered one, so
        // which constraint applies depends on the transport in play.
        auto supported_formats = router_->audio_speech_supported_formats(speech_model);
        if (is_streaming) {
            auto streaming_formats = router_->audio_speech_supported_streaming_formats(speech_model);
            if (!streaming_formats.empty()) {
                supported_formats = std::move(streaming_formats);
            }
        }
        // Streaming and encoding are orthogonal: stream_format picks the transport,
        // response_format the container. An explicit response_format therefore wins
        // even when streaming, and is rejected below rather than silently swapped if
        // the backend can't emit it. Only the implicit default differs by transport,
        // and it yields to whatever the backend declares.
        std::string response_format;
        if (request_json.contains("response_format") && request_json["response_format"].is_string()) {
            response_format = request_json["response_format"].get<std::string>();
        } else {
            response_format = is_streaming ? "pcm" : "mp3";
            const bool default_supported = supported_formats.empty() ||
                std::find(supported_formats.begin(), supported_formats.end(),
                          response_format) != supported_formats.end();
            if (!default_supported) {
                response_format = supported_formats.front();
            }
        }
        if (!MIME_TYPES.contains(response_format)) {
            res.status = 400;
            nlohmann::json error = {{"error", {
                {"message", "Unsupported audio format requested"},
                {"type", "invalid_request_error"}
            }}};
            res.set_content(error.dump(), "application/json");
            return;
        }
        const bool format_supported = supported_formats.empty() ||
            std::find(supported_formats.begin(), supported_formats.end(),
                      response_format) != supported_formats.end();
        // TODO: transcode from a natively supported format instead of rejecting.
        if (!format_supported) {
            std::string supported_list;
            for (const auto& f : supported_formats) {
                supported_list += (supported_list.empty() ? "" : ", ") + f;
            }
            res.status = 400;
            nlohmann::json error = {{"error", {
                {"message", "response_format '" + response_format + "' is not supported by this model "
                            "(supported: " + supported_list + ")"},
                {"type", "invalid_request_error"}
            }}};
            res.set_content(error.dump(), "application/json");
            return;
        }
        const auto format_metadata =
            router_->audio_speech_format_metadata(speech_model, response_format);
        std::string mime_type = format_metadata.content_type.empty()
            ? MIME_TYPES[response_format].get<std::string>()
            : format_metadata.content_type;
        for (const auto& [name, value] : format_metadata.headers) {
            res.set_header(name, value);
        }
        // The backend has to encode what the Content-Type above promises. Without
        // this it would receive whatever the client sent -- nothing, when the
        // format came from a default -- and fall back to its own choice.
        request_json["response_format"] = response_format;

        // Log the HTTP request
        LOG(INFO, "Server") << "POST /api/v1/audio/speech" << std::endl;

        auto audio_source = [this, request_json](size_t offset, httplib::DataSink& sink) {
            // For chunked responses, offset tracks bytes sent so far
            // We only want to stream once when offset is 0
            if (offset > 0) {
                return false; // We're done after the first call
            }

            // Use unified Router path for streaming
            router_->audio_speech(request_json, sink);

            return false;
        };

        if (is_streaming) {
            res.set_header("Cache-Control", "no-cache");
            res.set_header("Connection", "keep-alive");
            res.set_header("X-Accel-Buffering", "no"); // Disable nginx buffering

            // Use cpp-httplib's chunked content provider for streaming
            res.set_chunked_content_provider(mime_type, audio_source);
        } else {
            serve_media_or_error(res, mime_type, [this, request_json](httplib::DataSink& sink) {
                router_->audio_speech(request_json, sink);
            });
        }

        return;
    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "ERROR in handle_audio_speech: " << e.what() << std::endl;
        res.status = 500;
        nlohmann::json error = {{"error", {
            {"message", e.what()},
            {"type", "internal_error"}
        }}};
        res.set_content(error.dump(), "application/json");
    }
}

// The streaming plumbing reports backend failures as a payload in the sink —
// either an SSE-style "data: {\"error\":...}" event or the backend's own JSON
// error body — rather than throwing. Returns the parsed error, or null.
static nlohmann::json extract_error_payload(const std::string& buf) {
    std::string body = buf;
    if (body.rfind("data: ", 0) == 0) {
        body = body.substr(6);
    }
    const auto start = body.find_first_not_of(" \t\r\n");
    if (start == std::string::npos || body[start] != '{') {
        return nullptr;
    }
    try {
        auto parsed = nlohmann::json::parse(body.substr(start));
        if (parsed.is_object() && parsed.contains("error")) {
            return parsed;
        }
    } catch (...) {
    }
    return nullptr;
}

void Server::serve_media_or_error(httplib::Response& res, const std::string& mime_type,
                                  const std::function<void(httplib::DataSink&)>& generate) {
    // Buffer the generated media so we can tell success from a silent failure. A
    // streaming content provider commits a 200 before the backend runs, so a backend
    // crash/OOM mid-stream would surface as a successful empty file; buffering lets us
    // return a real error instead.
    std::string buf;
    httplib::DataSink sink;
    sink.write = [&buf](const char* data, size_t len) { buf.append(data, len); return true; };
    sink.is_writable = []() { return true; };
    sink.done = []() {};
    generate(sink);
    if (buf.empty()) {
        res.status = 502;
        res.set_content(nlohmann::json{{"error", {
            {"message", "Generation failed: the backend produced no output (it likely crashed or ran "
                        "out of GPU memory). Check the server logs."},
            {"type", "backend_error"}}}}.dump(), "application/json");
        return;
    }
    if (auto error_payload = extract_error_payload(buf); !error_payload.is_null()) {
        res.status = get_error_status_code(error_payload, 500);
        res.set_content(error_payload.dump(), "application/json");
        return;
    }
    res.set_content(buf, mime_type);
}

void Server::handle_audio_generations(const httplib::Request& req, httplib::Response& res) {
    try {
        auto request_json = nlohmann::json::parse(req.body);

        if (!request_json.contains("model")) {
            res.status = 400;
            res.set_content(nlohmann::json{{"error", {
                {"message", "Missing 'model' field in request"},
                {"type", "invalid_request_error"}}}}.dump(), "application/json");
            return;
        }
        if (!request_json.contains("prompt")) {
            res.status = 400;
            res.set_content(nlohmann::json{{"error", {
                {"message", "Missing 'prompt' field in request"},
                {"type", "invalid_request_error"}}}}.dump(), "application/json");
            return;
        }
        for (const auto* field : {"prompt", "lyrics", "vocal_language"}) {
            if (request_json.contains(field) && !request_json[field].is_string()) {
                res.status = 400;
                res.set_content(nlohmann::json{{"error", {
                    {"message", "'" + std::string(field) + "' must be a string"},
                    {"type", "invalid_request_error"}}}}.dump(), "application/json");
                return;
            }
        }

        std::string requested_model = request_json["model"];
        if (auto info = router_->try_get_model_info(requested_model);
            info && info->type != ModelType::AUDIO_GENERATION) {
            res.status = 400;
            res.set_content(nlohmann::json{{"error", {
                {"message", "Model '" + requested_model +
                    "' is not an audio-generation model"},
                {"type", "invalid_request_error"},
                {"code", "model_not_applicable"}}}}.dump(),
                "application/json");
            return;
        }
        try {
            auto_load_model_if_needed(requested_model, extract_auto_load_options(request_json));
        } catch (const std::exception& e) {
            LOG(ERROR, "Server") << "Failed to load audio-generation model: " << e.what() << std::endl;
            auto error_response = create_model_error(requested_model, e.what());
            res.status = get_http_status_from_error(error_response["error"]["code"].get<std::string>());
            res.set_content(error_response.dump(), "application/json");
            return;
        }

        std::string response_format = "wav";
        if (request_json.contains("response_format") && request_json["response_format"].is_string()) {
            response_format = request_json["response_format"].get<std::string>();
        }
        const auto supported_formats = router_->audio_generation_supported_formats(requested_model);
        const bool format_supported = std::find(supported_formats.begin(), supported_formats.end(),
                                                response_format) != supported_formats.end();
        // TODO: transcode from a natively supported format instead of rejecting.
        if (!supported_formats.empty() && !format_supported) {
            std::string supported_list;
            for (const auto& f : supported_formats) {
                supported_list += (supported_list.empty() ? "" : ", ") + f;
            }
            res.status = 400;
            res.set_content(nlohmann::json{{"error", {
                {"message", "response_format '" + response_format + "' is not supported by this model "
                            "(supported: " + supported_list + ")"},
                {"type", "invalid_request_error"}}}}.dump(), "application/json");
            return;
        }
        const auto format_metadata =
            router_->audio_generation_format_metadata(requested_model, response_format);
        std::string mime_type = !format_metadata.content_type.empty()
            ? format_metadata.content_type
            : (MIME_TYPES.contains(response_format)
                ? MIME_TYPES[response_format].get<std::string>()
                : MIME_TYPES["wav"].get<std::string>());
        for (const auto& [name, value] : format_metadata.headers) {
            res.set_header(name, value);
        }

        LOG(INFO, "Server") << "POST /api/v1/audio/generations" << std::endl;

        serve_media_or_error(res, mime_type, [this, request_json](httplib::DataSink& sink) {
            router_->audio_generations(request_json, sink);
        });
    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "ERROR in handle_audio_generations: " << e.what() << std::endl;
        res.status = 500;
        res.set_content(nlohmann::json{{"error", {
            {"message", e.what()}, {"type", "internal_error"}}}}.dump(), "application/json");
    }
}

void Server::handle_3d_generations(const httplib::Request& req, httplib::Response& res) {
    try {
        auto request_json = nlohmann::json::parse(req.body);

        // All cheap validation runs before auto_load_model_if_needed so a
        // malformed request can never trigger a multi-gigabyte model load.
        auto reject = [&res](const std::string& message) {
            res.status = 400;
            res.set_content(nlohmann::json{{"error", {
                {"message", message},
                {"type", "invalid_request_error"}}}}.dump(), "application/json");
        };
        if (!request_json.contains("model") || !request_json["model"].is_string()) {
            return reject("Missing or non-string 'model' field in request");
        }
        if (!request_json.contains("image") || !request_json["image"].is_string()) {
            return reject("Missing or non-string 'image' field in request (base64-encoded input image)");
        }
        {
            std::string image = request_json["image"].get<std::string>();
            if (image.rfind("data:", 0) == 0) {
                const auto comma = image.find(',');
                if (comma == std::string::npos) {
                    return reject("'image' data URL is missing its base64 payload");
                }
                image = image.substr(comma + 1);
            }
            if (image.empty() || image.find_first_not_of(
                    "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/=\r\n") !=
                    std::string::npos) {
                return reject("'image' must be base64 data or a base64 data: URL");
            }
            // Sniffing needs only the first 12 bytes; base64 is block-aligned,
            // so decoding the first 16 characters is enough and avoids decoding
            // a multi-megabyte payload just to validate it.
            std::string head;
            for (char c : image) {
                if (c != '\r' && c != '\n') {
                    head += c;
                    if (head.size() == 16) break;
                }
            }
            if (!utils::sniff_image(utils::JsonUtils::base64_decode(head)).ok()) {
                return reject("'image' is not a supported format (expected PNG, JPEG, BMP, or GIF)");
            }
        }
        std::string response_format = "glb";
        if (request_json.contains("response_format")) {
            if (!request_json["response_format"].is_string()) {
                return reject("'response_format' must be a string");
            }
            response_format = request_json["response_format"].get<std::string>();
        }
        // TODO: convert from a natively supported format instead of rejecting.
        if (response_format != "glb") {
            return reject("response_format '" + response_format +
                          "' is not supported (supported: glb)");
        }
        if (request_json.contains("resolution")) {
            const auto& r = request_json["resolution"];
            const std::string v = r.is_string() ? r.get<std::string>()
                : (r.is_number_integer() ? std::to_string(r.get<int>()) : "");
            if (v != "512" && v != "1024" && v != "1536") {
                return reject("'resolution' must be 512, 1024, or 1536");
            }
        }
        if (request_json.contains("bg_removal")) {
            const auto& b = request_json["bg_removal"];
            if (!b.is_string() || (b != "threshold" && b != "birefnet")) {
                return reject("'bg_removal' must be 'threshold' or 'birefnet'");
            }
        }
        if (request_json.contains("seed") && !request_json["seed"].is_number_integer()) {
            return reject("'seed' must be an integer");
        }
        if (request_json.contains("uv")) {
            const auto& u = request_json["uv"];
            if (!u.is_string() || (u != "box" && u != "xatlas")) {
                return reject("'uv' must be 'box' or 'xatlas'");
            }
        }

        std::string requested_model = request_json["model"];
        try {
            auto_load_model_if_needed(requested_model, extract_auto_load_options(request_json));
        } catch (const std::exception& e) {
            LOG(ERROR, "Server") << "Failed to load 3D-generation model: " << e.what() << std::endl;
            auto error_response = create_model_error(requested_model, e.what());
            res.status = get_http_status_from_error(error_response["error"]["code"].get<std::string>());
            res.set_content(error_response.dump(), "application/json");
            return;
        }

        LOG(INFO, "Server") << "POST /api/v1/3d/generations" << std::endl;

        serve_media_or_error(res, "model/gltf-binary", [this, request_json](httplib::DataSink& sink) {
            router_->model_3d_generations(request_json, sink);
        });
    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "ERROR in handle_3d_generations: " << e.what() << std::endl;
        res.status = 500;
        res.set_content(nlohmann::json{{"error", {
            {"message", e.what()}, {"type", "internal_error"}}}}.dump(), "application/json");
    }
}

void Server::handle_image_generations(const httplib::Request& req, httplib::Response& res) {
    try {
        LOG(INFO, "Server") << "POST /api/v1/images/generations" << std::endl;

        auto request_json = nlohmann::json::parse(req.body);
        normalize_client_model_name(request_json);
        normalize_and_resolve_request_model(request_json);

        // Validate required fields
        if (!request_json.contains("prompt")) {
            res.status = 400;
            nlohmann::json error = {{"error", {
                {"message", "Missing 'prompt' field in request"},
                {"type", "invalid_request_error"}
            }}};
            res.set_content(error.dump(), "application/json");
            return;
        }

        if (!request_json.contains("model")) {
            res.status = 400;
            nlohmann::json error = {{"error", {
                {"message", "Missing 'model' field in request"},
                {"type", "invalid_request_error"}
            }}};
            res.set_content(error.dump(), "application/json");
            return;
        }

        std::string requested_model = request_json["model"];

        try {
            auto_load_model_if_needed(requested_model, extract_auto_load_options(request_json));
        } catch (const std::exception& e) {
            LOG(ERROR, "Server") << "Failed to load image model: " << e.what() << std::endl;
            auto error_response = create_model_error(requested_model, e.what());
            std::string error_code = error_response["error"]["code"].get<std::string>();
            res.status = get_http_status_from_error(error_code);
            res.set_content(error_response.dump(), "application/json");
            return;
        }

        {
            auto response = router_->image_generations(request_json);
            if (response.contains("error")) {
                LOG(ERROR, "Server") << "Image generation backend error: " << response.dump() << std::endl;
                res.status = 500;
            }
            res.set_content(response.dump(), "application/json");
        }

    } catch (const nlohmann::json::exception& e) {
        LOG(ERROR, "Server") << "JSON parse error in handle_image_generations: " << e.what() << std::endl;
        res.status = 400;
        nlohmann::json error = {{"error", {
            {"message", "Invalid JSON: " + std::string(e.what())},
            {"type", "invalid_request_error"}
        }}};
        res.set_content(error.dump(), "application/json");
    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "ERROR in handle_image_generations: " << e.what() << std::endl;
        res.status = 500;
        nlohmann::json error = {{"error", {
            {"message", e.what()},
            {"type", "internal_error"}
        }}};
        res.set_content(error.dump(), "application/json");
    }
}

bool Server::parse_n_from_form(const httplib::Request& req, httplib::Response& res, nlohmann::json& out) {
    if (!req.form.has_field("n")) {
        return true;
    }
    int n;
    try {
        size_t pos;
        const std::string& val = req.form.get_field("n");
        n = std::stoi(val, &pos);
        if (pos != val.size()) throw std::invalid_argument("trailing characters");
    } catch (const std::exception&) {
        res.status = 400;
        nlohmann::json error = {{"error", {
            {"message", "Invalid value for 'n': must be an integer"},
            {"type", "invalid_request_error"}
        }}};
        res.set_content(error.dump(), "application/json");
        return false;
    }
    if (n < 1 || n > 10) {
        res.status = 400;
        nlohmann::json error = {{"error", {
            {"message", "Invalid value for 'n': must be between 1 and 10"},
            {"type", "invalid_request_error"}
        }}};
        res.set_content(error.dump(), "application/json");
        return false;
    }
    out["n"] = n;
    return true;
}

bool Server::extract_image_from_form(const httplib::Request& req, httplib::Response& res, nlohmann::json& out) {
    for (const auto& file_pair : req.form.files) {
        if (file_pair.first == "image" || file_pair.first == "image[]") {
            const auto& file = file_pair.second;
            out["image_data"] = utils::JsonUtils::base64_encode(file.content);
            out["image_filename"] = file.filename;
            LOG(INFO, "Server") << "Image file: " << file.filename
                      << " (" << file.content.size() << " bytes)" << std::endl;
            return true;
        }
    }
    res.status = 400;
    nlohmann::json error = {{"error", {
        {"message", "Missing 'image' field in request"},
        {"type", "invalid_request_error"}
    }}};
    res.set_content(error.dump(), "application/json");
    return false;
}

bool Server::load_image_model(const nlohmann::json& request_json, httplib::Response& res) {
    if (!request_json.contains("model")) {
        res.status = 400;
        nlohmann::json error = {{"error", {
            {"message", "Missing 'model' field in request"},
            {"type", "invalid_request_error"}
        }}};
        res.set_content(error.dump(), "application/json");
        return false;
    }
    std::string requested_model = request_json["model"];
    try {
        auto_load_model_if_needed(requested_model, extract_auto_load_options(request_json));
    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "Failed to load image model: " << e.what() << std::endl;
        auto error_response = create_model_error(requested_model, e.what());
        std::string error_code = error_response["error"]["code"].get<std::string>();
        res.status = get_http_status_from_error(error_code);
        res.set_content(error_response.dump(), "application/json");
        return false;
    }
    return true;
}

void Server::handle_image_edits(const httplib::Request& req, httplib::Response& res) {
    try {
        LOG(INFO, "Server") << "POST /api/v1/images/edits" << std::endl;

        if (!req.is_multipart_form_data()) {
            res.status = 400;
            nlohmann::json error = {{"error", {
                {"message", "Request must be multipart/form-data"},
                {"type", "invalid_request_error"}
            }}};
            res.set_content(error.dump(), "application/json");
            return;
        }

        nlohmann::json request_json;

        // Extract common form fields
        if (req.form.has_field("model"))            request_json["model"]            = req.form.get_field("model");
        if (req.form.has_field("prompt"))           request_json["prompt"]           = req.form.get_field("prompt");
        if (req.form.has_field("size"))             request_json["size"]             = req.form.get_field("size");
        if (req.form.has_field("response_format"))  request_json["response_format"]  = req.form.get_field("response_format");
        if (req.form.has_field("user"))             request_json["user"]             = req.form.get_field("user");
        if (req.form.has_field("background"))       request_json["background"]       = req.form.get_field("background");
        if (req.form.has_field("quality"))          request_json["quality"]          = req.form.get_field("quality");
        if (req.form.has_field("input_fidelity"))   request_json["input_fidelity"]   = req.form.get_field("input_fidelity");

        if (req.form.has_field("output_compression")) {
            int output_compression;
            try {
                size_t pos;
                const std::string& val = req.form.get_field("output_compression");
                output_compression = std::stoi(val, &pos);
                if (pos != val.size()) throw std::invalid_argument("trailing characters");
            } catch (const std::exception&) {
                res.status = 400;
                nlohmann::json error = {{"error", {
                    {"message", "Invalid value for 'output_compression': must be an integer"},
                    {"type", "invalid_request_error"}
                }}};
                res.set_content(error.dump(), "application/json");
                return;
            }
            request_json["output_compression"] = output_compression;
        }

        // Extract optional numeric inference parameters
        auto parse_int_field = [&](const std::string& field) -> bool {
            if (!req.form.has_field(field)) return true;
            const std::string& val = req.form.get_field(field);
            try {
                size_t pos;
                int parsed = std::stoi(val, &pos);
                if (pos != val.size()) throw std::invalid_argument("trailing characters");
                request_json[field] = parsed;
            } catch (const std::exception&) {
                res.status = 400;
                nlohmann::json error = {{"error", {
                    {"message", "Invalid value for '" + field + "': must be an integer"},
                    {"type", "invalid_request_error"}
                }}};
                res.set_content(error.dump(), "application/json");
                return false;
            }
            return true;
        };
        auto parse_float_field = [&](const std::string& field) -> bool {
            if (!req.form.has_field(field)) return true;
            const std::string& val = req.form.get_field(field);
            try {
                size_t pos;
                float parsed = std::stof(val, &pos);
                if (pos != val.size()) throw std::invalid_argument("trailing characters");
                if (std::isnan(parsed) || std::isinf(parsed)) throw std::invalid_argument("nan/inf not allowed");
                request_json[field] = parsed;
            } catch (const std::exception&) {
                res.status = 400;
                nlohmann::json error = {{"error", {
                    {"message", "Invalid value for '" + field + "': must be a number"},
                    {"type", "invalid_request_error"}
                }}};
                res.set_content(error.dump(), "application/json");
                return false;
            }
            return true;
        };
        if (!parse_int_field("steps"))     return;
        if (!parse_float_field("cfg_scale")) return;
        if (!parse_int_field("seed"))      return;

        if (!parse_n_from_form(req, res, request_json))      return;
        if (!extract_image_from_form(req, res, request_json)) return;

        // Extract optional mask file
        for (const auto& file_pair : req.form.files) {
            if (file_pair.first == "mask") {
                const auto& file = file_pair.second;
                request_json["mask_data"] = utils::JsonUtils::base64_encode(file.content);
                request_json["mask_filename"] = file.filename;
                LOG(INFO, "Server") << "Mask file: " << file.filename
                          << " (" << file.content.size() << " bytes)" << std::endl;
                break;
            }
        }

        if (!request_json.contains("prompt")) {
            res.status = 400;
            nlohmann::json error = {{"error", {
                {"message", "Missing 'prompt' field in request"},
                {"type", "invalid_request_error"}
            }}};
            res.set_content(error.dump(), "application/json");
            return;
        }

        if (!load_image_model(request_json, res)) return;

        auto response = router_->image_edits(request_json);
        if (response.contains("error")) {
            LOG(ERROR, "Server") << "Image edits backend error: " << response.dump() << std::endl;
            res.status = 500;
        }
        res.set_content(response.dump(), "application/json");

    } catch (const nlohmann::json::exception& e) {
        LOG(ERROR, "Server") << "JSON parse error in handle_image_edits: " << e.what() << std::endl;
        res.status = 400;
        nlohmann::json error = {{"error", {
            {"message", "Invalid JSON: " + std::string(e.what())},
            {"type", "invalid_request_error"}
        }}};
        res.set_content(error.dump(), "application/json");
    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "ERROR in handle_image_edits: " << e.what() << std::endl;
        res.status = 500;
        nlohmann::json error = {{"error", {
            {"message", e.what()},
            {"type", "server_error"}
        }}};
        res.set_content(error.dump(), "application/json");
    }
}

void Server::handle_image_variations(const httplib::Request& req, httplib::Response& res) {
    try {
        LOG(INFO, "Server") << "POST /api/v1/images/variations" << std::endl;

        if (!req.is_multipart_form_data()) {
            res.status = 400;
            nlohmann::json error = {{"error", {
                {"message", "Request must be multipart/form-data"},
                {"type", "invalid_request_error"}
            }}};
            res.set_content(error.dump(), "application/json");
            return;
        }

        nlohmann::json request_json;

        // Extract common form fields
        if (req.form.has_field("model"))            request_json["model"]            = req.form.get_field("model");
        if (req.form.has_field("size"))             request_json["size"]             = req.form.get_field("size");
        if (req.form.has_field("response_format"))  request_json["response_format"]  = req.form.get_field("response_format");
        if (req.form.has_field("user"))             request_json["user"]             = req.form.get_field("user");

        if (!parse_n_from_form(req, res, request_json))      return;
        if (!extract_image_from_form(req, res, request_json)) return;
        if (!load_image_model(request_json, res))             return;

        auto response = router_->image_variations(request_json);
        if (response.contains("error")) {
            LOG(ERROR, "Server") << "Image variations backend error: " << response.dump() << std::endl;
            res.status = 500;
        }
        res.set_content(response.dump(), "application/json");

    } catch (const nlohmann::json::exception& e) {
        LOG(ERROR, "Server") << "JSON parse error in handle_image_variations: " << e.what() << std::endl;
        res.status = 400;
        nlohmann::json error = {{"error", {
            {"message", "Invalid JSON: " + std::string(e.what())},
            {"type", "invalid_request_error"}
        }}};
        res.set_content(error.dump(), "application/json");
    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "ERROR in handle_image_variations: " << e.what() << std::endl;
        res.status = 500;
        nlohmann::json error = {{"error", {
            {"message", e.what()},
            {"type", "server_error"}
        }}};
        res.set_content(error.dump(), "application/json");
    }
}

void Server::handle_image_upscale(const httplib::Request& req, httplib::Response& res) {
    try {
        LOG(INFO, "Server") << "POST /api/v1/images/upscale" << std::endl;

        auto request_json = nlohmann::json::parse(req.body);

        if (!request_json.contains("image") || !request_json["image"].is_string()) {
            res.status = 400;
            nlohmann::json error = {{"error", {
                {"message", "Missing 'image' field (base64 encoded)"},
                {"type", "invalid_request_error"}
            }}};
            res.set_content(error.dump(), "application/json");
            return;
        }

        std::string upscale_model_name = request_json.value("model", "");
        if (upscale_model_name.empty()) {
            res.status = 400;
            nlohmann::json error = {{"error", {
                {"message", "Missing 'model' field"},
                {"type", "invalid_request_error"}
            }}};
            res.set_content(error.dump(), "application/json");
            return;
        }

        std::string upscale_model_path;
        std::string recipe;
        try {
            auto info = model_manager_->get_model_info(upscale_model_name);

            if (!model_manager_->is_model_downloaded(upscale_model_name)) {
                LOG(INFO, "Server") << "Upscale model not cached, downloading from its remote registry..." << std::endl;
                model_manager_->download_registered_model(info, true);
                LOG(INFO, "Server") << "Upscale model download complete: " << upscale_model_name << std::endl;
                info = model_manager_->get_model_info(upscale_model_name);
            }

            upscale_model_path = info.resolved_path("main");
            recipe = info.recipe;
        } catch (const std::exception& e) {
            res.status = 404;
            nlohmann::json error = {{"error", {
                {"message", "Upscale model not found: " + upscale_model_name},
                {"type", "invalid_request_error"}
            }}};
            res.set_content(error.dump(), "application/json");
            return;
        }

        std::string b64_image = request_json["image"].get<std::string>();

        // Upscaling is model-free: no model is loaded through the router, so we
        // dispatch by recipe to each backend's shared upscale API, which shells
        // out to its own CLI binary directly (backend selection, binary path,
        // and runtime environment all live in the backend).
        std::string upscaled;
        if (recipe == "thenoise") {
            upscaled = lemon::backends::TheNoiseServer::upscale_via_cli(b64_image, upscale_model_path);
        } else if (recipe == "sd-cpp") {
            upscaled = lemon::backends::SDServer::upscale_via_cli(b64_image, upscale_model_path);
        } else {
            res.status = 400;
            nlohmann::json error = {{"error", {
                {"message", "Upscale is not supported by recipe: " + recipe},
                {"type", "invalid_request_error"}
            }}};
            res.set_content(error.dump(), "application/json");
            return;
        }

        if (upscaled.empty()) {
            res.status = 500;
            nlohmann::json error = {{"error", {
                {"message", "Upscale failed"},
                {"type", "server_error"}
            }}};
            res.set_content(error.dump(), "application/json");
            return;
        }

        nlohmann::json response;
        response["created"] = static_cast<long long>(std::time(nullptr));
        response["data"] = nlohmann::json::array();
        response["data"].push_back({{"b64_json", upscaled}});
        res.set_content(response.dump(), "application/json");

    } catch (const nlohmann::json::exception& e) {
        LOG(ERROR, "Server") << "JSON parse error in handle_image_upscale: " << e.what() << std::endl;
        res.status = 400;
        nlohmann::json error = {{"error", {
            {"message", "Invalid JSON: " + std::string(e.what())},
            {"type", "invalid_request_error"}
        }}};
        res.set_content(error.dump(), "application/json");
    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "ERROR in handle_image_upscale: " << e.what() << std::endl;
        res.status = 500;
        nlohmann::json error = {{"error", {
            {"message", e.what()},
            {"type", "server_error"}
        }}};
        res.set_content(error.dump(), "application/json");
    }
}

void Server::handle_responses(const httplib::Request& req, httplib::Response& res) {
    try {
        auto request_json = nlohmann::json::parse(req.body);
        normalize_client_model_name(request_json);
        normalize_and_resolve_request_model(request_json);

        // A collection.router model flips this endpoint into engine mode: pick a
        // candidate and rewrite the model before the usual load/forward logic.
        std::optional<RouterDispatchResult> route_dispatch =
            apply_router_collection_dispatch(request_json);

        // Handle model loading/switching using helper function
        if (request_json.contains("model")) {
            std::string requested_model = request_json["model"];
            try {
                auto_load_model_if_needed(requested_model, extract_auto_load_options(request_json));
            } catch (const std::exception& e) {
                LOG(ERROR, "Server") << "Failed to load model: " << e.what() << std::endl;
                auto error_response = create_model_error(requested_model, e.what());
                std::string error_code = error_response["error"]["code"].get<std::string>();
                res.status = get_http_status_from_error(error_code);
                res.set_content(error_response.dump(), "application/json");
                return;
            }
        } else if (!router_->is_model_loaded()) {
            LOG(ERROR, "Server") << "No model loaded and no model specified in request" << std::endl;
            res.status = 400;
            res.set_content("{\"error\": \"No model loaded and no model specified in request\"}", "application/json");
            return;
        }

        // Check if streaming is requested
        bool is_streaming = request_json.contains("stream") && request_json["stream"].get<bool>();

        if (is_streaming) {
            try {
                // Re-serialize so any collection.router model rewrite reaches the backend.
                std::string request_body = request_json.dump();
                LOG(INFO, "Server") << "POST /api/v1/responses - Streaming" << std::endl;

                set_route_decision_sse_content_provider(
                    res,
                    route_dispatch,
                    request_body,
                    [this](const std::string& body, httplib::DataSink& sink) {
                        router_->responses_stream(body, sink);
                    });
            } catch (const std::exception& e) {
                LOG(ERROR, "Server") << "Streaming failed: " << e.what() << std::endl;
                res.status = 500;
                res.set_content("{\"error\":\"Internal server error during streaming\"}", "application/json");
            }
        } else {
            LOG(INFO, "Server") << "POST /api/v1/responses - Non-streaming" << std::endl;

            auto response = router_->responses(request_json);

            if (response.contains("error")) {
                LOG(ERROR, "Server") << "Responses backend error: " << response["error"].dump() << std::endl;
                set_error_response(response, res);
                return;
            }

            attach_route_decision(response, res, route_dispatch);
            LOG(INFO, "Server") << "200 OK" << std::endl;
            res.set_content(response.dump(), "application/json");

            record_response_telemetry(response, request_json);
        }

    } catch (const RouterResidencyConflictException& e) {
        LOG(WARNING, "Server") << "Router residency conflict in responses: "
                               << e.what() << std::endl;
        set_router_residency_conflict_response(e, res);
    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "ERROR in handle_responses: " << e.what() << std::endl;
        res.status = 500;
        nlohmann::json error = {{"error", e.what()}};
        res.set_content(error.dump(), "application/json");
    }
}
bool Server::parse_required_json_body(const httplib::Request& req,
                                      httplib::Response& res,
                                      nlohmann::json& out) {
    if (req.body.empty()) {
        res.status = 400;
        nlohmann::json error = {{"error", "Request body is required but was empty"}};
        res.set_content(error.dump(), "application/json");
        return false;
    }
    try {
        out = nlohmann::json::parse(req.body);
        return true;
    } catch (const nlohmann::json::exception& e) {
        // Not just parse_error: a numeric literal too large for a double raises
        // out_of_range, which is a sibling type and would otherwise escape as a 500.
        res.status = 400;
        nlohmann::json error = {{"error", std::string("Invalid JSON in request body: ") + e.what()}};
        res.set_content(error.dump(), "application/json");
        return false;
    }
}

void Server::handle_pull(const httplib::Request& req, httplib::Response& res) {
    auto bad_request = [&res](const std::string& message) {
        res.status = 400;
        nlohmann::json error = {{"error", message}};
        res.set_content(error.dump(), "application/json");
    };

    nlohmann::json request_json;
    if (!parse_required_json_body(req, res, request_json)) return;

    try {
        // Accept both "model" and "model_name" for compatibility
        std::string model_name = request_json.contains("model") ?
            request_json["model"].get<std::string>() :
            request_json["model_name"].get<std::string>();

        // Extract optional parameters
        std::string checkpoint = request_json.value("checkpoint", "");
        std::string recipe = request_json.value("recipe", "");
        bool do_not_upgrade = request_json.value("do_not_upgrade", false);
        bool stream = request_json.value("stream", false);
        bool subscribe = request_json.value("subscribe", true);
        bool local_import = request_json.value("local_import", false);

        // Resolve the pull's registry provenance once, up front, so download,
        // cache layout, and later refresh stay consistent. A registry-backed
        // checkpoint that names no source inherits the configured default; a
        // provider URL is normalized to owner/repo and its registry adopted.
        // Self-managed backends (flm/cloud), non-registry checkpoints, explicit
        // sources, and bare `pull <registered-name>` refreshes are left as-is.
        try {
            lemon::apply_default_pull_source(request_json, config_->default_model_source());
        } catch (const std::invalid_argument& e) {
            bad_request(e.what());
            return;
        }
        // The helper may have rewritten a provider URL into an owner/repo id.
        checkpoint = request_json.value("checkpoint", "");

        LOG(INFO, "Server") << "Pulling model: " << model_name << std::endl;
        if (!checkpoint.empty()) {
            LOG(INFO, "Server") << "   checkpoint: " << checkpoint << std::endl;
        }
        if (!recipe.empty()) {
            LOG(INFO, "Server") << "   recipe: " << recipe << std::endl;
        }

        // Both API operations always enter the same registration path.
        // /pull continues with installation after this; /models/register returns.
        const bool collection_file_import =
            request_json.contains("models") && request_json["models"].is_array();

        try {
            register_model_definition_internal(
                model_name,
                request_json,
                /*require_definition=*/false,
                /*allow_embedded_models=*/true,
                local_import);
        } catch (const std::invalid_argument& e) {
            bad_request(e.what());
            return;
        }

        if (local_import) {
            nlohmann::json response = {
                {"status", "success"},
                {"model_name", model_name},
                {"message", "Model imported and registered successfully"}
            };
            res.set_content(response.dump(), "application/json");
            return;
        }

        if (config_->offline()) {
            res.status = 400;
            nlohmann::json error = {{"error", "Lemond is in offline mode, models not downloaded"}, {"code", "lemond_offline"}};
            res.set_content(error.dump(), "application/json");
            return;
        }

        nlohmann::json download_request = collection_file_import
            ? request_json
            : nlohmann::json::object();

        if (stream) {
            auto operation = [this, model_name, download_request, do_not_upgrade](DownloadProgressCallback progress_cb) {
                model_manager_->download_model(model_name, download_request, do_not_upgrade, progress_cb);
            };

            if (!subscribe) {
                // Server-owned mode for desktop UI reload/new-tab recovery.
                // Legacy streamed /pull behavior is unchanged.
                auto job = start_download_job("model:" + model_name, "model", model_name, operation);
                nlohmann::json response;
                {
                    // The worker can update the shared job immediately after
                    // start_download_job returns, so copy the response while
                    // holding the same mutex used by the progress callback.
                    std::lock_guard<std::mutex> lock(downloads_mutex_);
                    response = download_job_to_json(job);
                }
                res.set_content(response.dump(), "application/json");
                return;
            }

            // Backward-compatible SSE streaming mode: tie the operation to this
            // response exactly as before.
            stream_download_operation(res, operation);
        } else {
            // Legacy synchronous mode - blocks until complete
            model_manager_->download_model(model_name, download_request, do_not_upgrade);

            nlohmann::json response = {{"status", "success"}, {"model_name", model_name}};
            res.set_content(response.dump(), "application/json");
        }

    } catch (const lemon::UnknownModelError& e) {
        LOG(ERROR, "Server") << "ERROR in handle_pull: " << e.what() << std::endl;
        res.status = 400;
        nlohmann::json error = {{"error", e.what()}, {"code", lemon::kUnknownModelErrorCode}};
        res.set_content(error.dump(), "application/json");
    } catch (const lemon::InvalidModelDefinitionError& e) {
        LOG(ERROR, "Server") << "ERROR in handle_pull: " << e.what() << std::endl;
        res.status = 400;
        nlohmann::json error = {{"error", e.what()}};
        res.set_content(error.dump(), "application/json");
    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "ERROR in handle_pull: " << e.what() << std::endl;
        res.status = 500;
        nlohmann::json error = {{"error", e.what()}};
        res.set_content(error.dump(), "application/json");
    }
}

void Server::handle_registry_search(const httplib::Request& req, httplib::Response& res) {
    try {
        std::string query = req.has_param("query")
            ? req.get_param_value("query")
            : (req.has_param("q") ? req.get_param_value("q") : "");
        const auto first = query.find_first_not_of(" \t\r\n");
        const auto last = query.find_last_not_of(" \t\r\n");
        query = first == std::string::npos ? "" : query.substr(first, last - first + 1);
        if (query.size() < 3) {
            res.status = 400;
            res.set_content(nlohmann::json{{"error", {
                {"message", "Query parameter 'query' must contain at least 3 characters"},
                {"type", "invalid_request_error"}
            }}}.dump(), "application/json");
            return;
        }

        const std::string source_text = req.has_param("source")
            ? req.get_param_value("source") : "huggingface";
        const auto source = parse_remote_registry_source(source_text);

        std::size_t limit = 12;
        if (req.has_param("limit")) {
            const std::string limit_text = req.get_param_value("limit");
            std::size_t parsed = 0;
            try {
                std::size_t consumed = 0;
                parsed = static_cast<std::size_t>(std::stoul(limit_text, &consumed));
                if (consumed != limit_text.size()) throw std::invalid_argument("trailing characters");
            } catch (...) {
                throw std::invalid_argument("Query parameter 'limit' must be an integer from 1 to 50");
            }
            if (parsed < 1 || parsed > 50) {
                throw std::invalid_argument("Query parameter 'limit' must be an integer from 1 to 50");
            }
            limit = parsed;
        }

        bool gguf_only = false;
        if (req.has_param("format")) {
            const std::string format = req.get_param_value("format");
            if (format != "gguf") {
                throw std::invalid_argument(
                    "Unsupported registry search format '" + format + "' (expected 'gguf')");
            }
            gguf_only = true;
        }

        if (config_->offline()) {
            res.status = 400;
            res.set_content(nlohmann::json{{"error", {
                {"message", "Lemond is in offline mode, registry search is unavailable"},
                {"type", "invalid_request_error"},
                {"code", "lemond_offline"}
            }}}.dump(), "application/json");
            return;
        }

        const auto search = search_registry_models(source, query, limit, gguf_only);
        nlohmann::json results = nlohmann::json::array();
        for (const auto& model : search.results) {
            results.push_back({
                {"repository_id", model.repo_id},
                {"display_name", model.display_name},
                {"source", remote_registry_source_name(model.source)},
                {"repository_type", model.repository_type},
                {"description", model.description},
                {"tags", model.tags},
                {"task", model.task},
                {"downloads", model.downloads},
                {"likes", model.likes},
                {"has_gguf", model.has_gguf}
            });
        }
        nlohmann::json response = {
            {"source", remote_registry_source_name(source)},
            {"query", query},
            {"total", search.total},
            {"results", std::move(results)}
        };
        if (gguf_only) response["format"] = "gguf";
        res.set_content(response.dump(), "application/json");
    } catch (const RegistrySearchError& e) {
        res.status = e.status_code() == 429 ? 429 : 502;
        res.set_content(nlohmann::json{{"error", {
            {"message", e.what()},
            {"type", "registry_error"},
            {"upstream_status_code", e.status_code()}
        }}}.dump(), "application/json");
    } catch (const std::invalid_argument& e) {
        res.status = 400;
        res.set_content(nlohmann::json{{"error", {
            {"message", e.what()},
            {"type", "invalid_request_error"}
        }}}.dump(), "application/json");
    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "ERROR in handle_registry_search: " << e.what() << std::endl;
        res.status = 502;
        res.set_content(nlohmann::json{{"error", {
            {"message", e.what()},
            {"type", "registry_error"}
        }}}.dump(), "application/json");
    }
}

void Server::handle_pull_variants(const httplib::Request& req, httplib::Response& res) {
    auto bad_request = [&res](const std::string& message) {
        res.status = 400;
        nlohmann::json error = {{"error", message}};
        res.set_content(error.dump(), "application/json");
    };
    try {
        std::string checkpoint = req.get_param_value("checkpoint");
        if (checkpoint.empty()) {
            res.status = 400;
            nlohmann::json error = {{"error", "Missing required query parameter 'checkpoint'"}};
            res.set_content(error.dump(), "application/json");
            return;
        }
        // Detect the URL provider upfront so we can reject mismatches with
        // an explicit `--source` param (just like /pull and the CLI do).
        lemon::RemoteRegistrySource url_source;
        std::string url_repo_id;
        bool has_url = lemon::detect_registry_url(checkpoint, url_source, url_repo_id);

        // Explicit `--source` was provided (or not).  Capture the raw param
        // now while checkpoint still holds the URL for error messages.
        std::string raw_source;
        if (req.has_param("source")) {
            raw_source = req.get_param_value("source");
        }

        // Reject mismatches between explicit source and detected URL source.
        // Canonicalize the raw param first so accepted aliases (e.g. `hf`) are
        // not rejected against the URL's canonical registry name.
        if (has_url && !raw_source.empty()) {
            if (parse_remote_registry_source(raw_source) != url_source) {
                bad_request(
                    "checkpoint URL uses " + remote_registry_source_name(url_source) +
                    " but source was set to '" + raw_source + "'");
                return;
            }
        }

        // A provider URL is authoritative: normalize it to owner/repo and adopt
        // its registry. Otherwise honor an explicit source param, falling back
        // to the server's configured default when none is given.
        if (has_url) {
            checkpoint = url_repo_id;
        }
        std::string source;
        if (has_url) {
            source = remote_registry_source_name(url_source);
        } else {
            source = req.has_param("source")
                ? req.get_param_value("source") : config_->default_model_source();
        }
        if (checkpoint.find('/') == std::string::npos) {
            res.status = 400;
            nlohmann::json error = {{"error",
                "Malformed 'checkpoint': expected a repository id of the form 'owner/name'"}};
            res.set_content(error.dump(), "application/json");
            return;
        }
        const auto parsed_source = parse_remote_registry_source(source);
        if (config_->offline()) {
            res.status = 400;
            nlohmann::json error = {{"error", "Lemond is in offline mode, models not downloaded"}, {"code", "lemond_offline"}};
            res.set_content(error.dump(), "application/json");
            return;
        }
        bool not_found = false;
        const std::string resolved_source = remote_registry_source_name(parsed_source);
        nlohmann::json body = lemon::fetch_pull_variants(
            checkpoint, resolved_source, not_found);
        if (not_found) {
            res.status = 404;
            nlohmann::json error = {{"error", "Checkpoint '" + checkpoint + "' not found on " +
                remote_registry_display_name(parsed_source)}};
            res.set_content(error.dump(), "application/json");
            return;
        }
        if (body.is_object()) {
            body["source"] = resolved_source;
        }
        res.set_content(body.dump(), "application/json");
    } catch (const std::invalid_argument& e) {
        res.status = 400;
        nlohmann::json error = {{"error", e.what()}};
        res.set_content(error.dump(), "application/json");
    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "ERROR in handle_pull_variants: " << e.what() << std::endl;
        res.status = 500;
        nlohmann::json error = {{"error", e.what()}};
        res.set_content(error.dump(), "application/json");
    }
}

void Server::handle_load(const httplib::Request& req, httplib::Response& res) {
    auto thread_id = std::this_thread::get_id();
    LOG(DEBUG, "Server") << "===== LOAD ENDPOINT ENTERED (Thread: " << thread_id << ") =====" << std::endl;

    // Declare model_name outside try block so it's available in catch block
    std::string model_name;

    nlohmann::json request_json;
    if (!parse_required_json_body(req, res, request_json)) return;
    normalize_and_resolve_request_model(request_json);

    try {
        model_name = request_json["model_name"];

        // Cloud models are registered automatically at cache build / cloud-auth
        // / install time, so a cloud model reaching /load is already in the
        // cache. /load carries no creds payload; CloudServer reads the
        // resolved key (env var or runtime POST) from CloudProviderRegistry
        // when a request is actually forwarded.

        // Get model info
        if (!model_manager_->model_exists(model_name)) {
            LOG(ERROR, "Server") << "Model not found: " << model_name << std::endl;
            res.status = 404;
            auto error_response = create_model_error(model_name, "Model not found");
            res.set_content(error_response.dump(), "application/json");
            return;
        }

        auto info = model_manager_->get_model_info(model_name);

        // Extract optional per-model settings. Omitted options keep saved values;
        // null masks only its saved key for this load. Concrete *_args scope is
        // resolved later by Router. ctx_size=-1 remains an explicit auto value.
        RecipeOptions options = RecipeOptions(info.recipe, request_json);
        std::set<std::string> transient_saved_option_tombstones;
        for (const auto& key : RecipeOptions::keys_for_recipe(info.recipe)) {
            if (request_json.contains(key) && request_json[key].is_null()) {
                transient_saved_option_tombstones.insert(key);
            }
        }

        bool save_options = request_json.value("save_options", false);
        std::optional<bool> pinned_opt = std::nullopt;
        if (request_json.contains("pinned") && request_json["pinned"].is_boolean()) {
            pinned_opt = request_json["pinned"].get<bool>();
        }

        LOG(INFO, "Server") << "Ensuring model loaded: " << model_name;
        LOG(INFO, "Server") << " " << options.to_log_string(false);
        LOG(INFO, "Server") << std::endl;

        // Persist concrete request options if requested. A null tombstone is
        // load-scoped, so preserve the existing saved value for that key.
        if (save_options) {
            json saved_for_write = options.to_json();
            if (!transient_saved_option_tombstones.empty()) {
                const json existing_saved = model_manager_->get_saved_model_options(model_name);
                for (const auto& key : transient_saved_option_tombstones) {
                    auto it = existing_saved.find(key);
                    if (it != existing_saved.end()) {
                        saved_for_write[key] = *it;
                    }
                }
            }
            model_manager_->set_saved_model_options(model_name, saved_for_write);
            info = model_manager_->get_model_info(model_name);
        }

        if (!model_manager_->is_model_downloaded(model_name) && !is_model_collection_recipe(info.recipe)) {
            LOG(INFO, "Server") << "Model not downloaded, downloading..." << std::endl;
            model_manager_->download_registered_model(info);
            info = model_manager_->get_model_info(model_name);
        }

        // Null tombstones are load-local: remove only those saved keys from the
        // local model layer without changing recipe_options.json.
        if (!transient_saved_option_tombstones.empty()) {
            json per_model_options = model_manager_->get_model_default_options(info).to_json();
            const json saved = model_manager_->get_saved_model_options(model_name);
            if (saved.is_object()) {
                for (const auto& [key, value] : saved.items()) {
                    if (transient_saved_option_tombstones.count(key) == 0) {
                        per_model_options[key] = value;
                    }
                }
            }
            info.recipe_options = RecipeOptions(info.recipe, per_model_options);
        }


        // Collection models: load each component instead
        if (is_omni_collection_recipe(info.recipe) && !info.components.empty()) {
            ensure_collection_loaded(info);

            nlohmann::json response = {
                {"status", "success"},
                {"model_name", model_name},
                {"recipe", info.recipe}
            };
            res.set_content(response.dump(), "application/json");
        } else if (is_router_collection_recipe(info.recipe)) {
            // Router collections are virtual: each request is dispatched to one
            // candidate at request time, which lazy-loads it. There is no backend
            // of the collection's own to bring up, and eagerly loading every
            // candidate would thrash the model LRU. Acknowledge without loading.
            nlohmann::json response = {
                {"status", "success"},
                {"model_name", model_name},
                {"recipe", info.recipe}
            };
            res.set_content(response.dump(), "application/json");
        } else {
            // Load model with optional per-model settings (declarative: no-op
            // if already loaded with matching options, reload only if options
            // differ)
            router_->load_model(model_name, info, options, true,
                                /*allow_reload_on_option_change=*/true,
                                pinned_opt);

            // Return success response
            nlohmann::json response = {
                {"status", "success"},
                {"model_name", model_name},
                {"checkpoint", info.checkpoint()},
                {"recipe", info.recipe}
            };
            res.set_content(response.dump(), "application/json");
        }

    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "Failed to load model: " << e.what() << std::endl;

        // Use consistent error format
        if (!model_name.empty()) {
            auto error_response = create_model_error(model_name, e.what());
            std::string error_code = error_response["error"]["code"].get<std::string>();
            res.status = get_http_status_from_error(error_code);
            res.set_content(error_response.dump(), "application/json");
        } else {
            // JSON parsing failed before we got model_name - return generic error
            res.status = 400;
            nlohmann::json error = {{"error", {
                {"message", std::string("Invalid request: ") + e.what()},
                {"type", "invalid_request_error"},
                {"code", "invalid_request"}
            }}};
            res.set_content(error.dump(), "application/json");
        }
    }
}

void Server::handle_load_command(const httplib::Request& req, httplib::Response& res) {
    nlohmann::json request_json;
    if (!parse_required_json_body(req, res, request_json)) return;

    std::string model_name;
    try {
        model_name = request_json.value("model_name", "");
        if (model_name.empty()) {
            res.status = 400;
            res.set_content(nlohmann::json{{"error", "model_name is required"}}.dump(), "application/json");
            return;
        }

        if (!model_manager_->model_exists(model_name)) {
            res.status = 404;
            res.set_content(create_model_error(model_name, "Model not found").dump(), "application/json");
            return;
        }

        auto info = model_manager_->get_model_info(model_name);
        RecipeOptions options = RecipeOptions(info.recipe, request_json);
        RecipeOptions effective = router_->resolve_effective_recipe_options(info, options);

        bool ctx_size_auto_resolved = false;
        int64_t auto_ctx = resolve_auto_ctx_size(effective, info);
        if (auto_ctx > 0) {
            effective.set_option("ctx_size", auto_ctx);
            ctx_size_auto_resolved = true;
        }

        json backend_json = effective.get_option(info.recipe + "_backend");

        nlohmann::json response = {
            {"model_name", model_name},
            {"recipe", info.recipe},
            {"backend", backend_json.is_string() ? backend_json.get<std::string>() : ""},
            {"options", effective.to_json()},
            {"args", RecipeOptions::to_cli_options(effective.to_json())},
            {"ctx_size_auto_resolved", ctx_size_auto_resolved}
        };
        res.set_content(response.dump(), "application/json");
    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "ERROR in handle_load_command: " << e.what() << std::endl;
        res.status = 500;
        auto error_response = model_name.empty()
            ? nlohmann::json{{"error", e.what()}}
            : create_model_error(model_name, e.what());
        res.set_content(error_response.dump(), "application/json");
    }
}

void Server::handle_unload(const httplib::Request& req, httplib::Response& res) {
    try {
        LOG(INFO, "Server") << "Unload request received" << std::endl;
        LOG(DEBUG, "Server") << "Request method: " << req.method << ", body length: " << req.body.length() << std::endl;
        LOG(DEBUG, "Server") << "Content-Type: " << req.get_header_value("Content-Type") << std::endl;

        // Multi-model support: Optional model_name parameter
        std::string model_name;
        if (!req.body.empty()) {
            try {
                auto request_json = nlohmann::json::parse(req.body);
                normalize_client_model_name(request_json);
                normalize_and_resolve_request_model(request_json);
                if (request_json.contains("model_name") && request_json["model_name"].is_string()) {
                    model_name = request_json["model_name"].get<std::string>();
                } else if (request_json.contains("model") && request_json["model"].is_string()) {
                    model_name = request_json["model"].get<std::string>();
                }
            } catch (...) {
                // Ignore parse errors, just unload all
            }
        }

        router_->unload_model(model_name);  // Empty string = unload all

        if (model_name.empty()) {
            LOG(INFO, "Server") << "All models unloaded successfully" << std::endl;
            nlohmann::json response = {
                {"status", "success"},
                {"message", "All models unloaded successfully"}
            };
            res.status = 200;
            res.set_content(response.dump(), "application/json");
        } else {
            LOG(INFO, "Server") << "Model '" << model_name << "' unloaded successfully" << std::endl;
            nlohmann::json response = {
                {"status", "success"},
                {"message", "Model unloaded successfully"},
                {"model_name", model_name}
            };
            res.status = 200;
            res.set_content(response.dump(), "application/json");
        }
    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "Unload failed: " << e.what() << std::endl;

        // Check if error is "Model not loaded" for 404
        std::string error_msg = e.what();
        if (error_msg.find("not loaded") != std::string::npos) {
            res.status = 404;
        } else {
            res.status = 500;
        }

        nlohmann::json error = {{"error", e.what()}};
        res.set_content(error.dump(), "application/json");
    }
}

void Server::handle_pin(const httplib::Request& req, httplib::Response& res) {
    try {
        nlohmann::json request_json;
        try {
            request_json = nlohmann::json::parse(req.body);
        } catch (const std::exception& parse_err) {
            res.status = 400;
            res.set_content(nlohmann::json{{"error", {
                {"message", "Invalid JSON body: " + std::string(parse_err.what())},
                {"type", "invalid_request_error"}
            }}}.dump(), "application/json");
            return;
        }

        if (!request_json.contains("model") && !request_json.contains("model_name")) {
            res.status = 400;
            res.set_content(nlohmann::json{{"error", {
                {"message", "Parameter 'model' or 'model_name' is required"},
                {"type", "invalid_request_error"}
            }}}.dump(), "application/json");
            return;
        }

        if (!request_json.contains("pinned") || !request_json["pinned"].is_boolean()) {
            res.status = 400;
            res.set_content(nlohmann::json{{"error", {
                {"message", "Parameter 'pinned' is required and must be a boolean"},
                {"type", "invalid_request_error"}
            }}}.dump(), "application/json");
            return;
        }

        std::string model_name = request_json.contains("model") ?
            request_json["model"].get<std::string>() :
            request_json["model_name"].get<std::string>();
        bool pinned = request_json["pinned"].get<bool>();

        std::string canonical_name = model_manager_->resolve_model_name(model_name);
        router_->set_model_pinned(canonical_name, pinned);

        nlohmann::json response = {
            {"status", "success"},
            {"model_name", model_name},
            {"pinned", pinned}
        };
        res.status = 200;
        res.set_content(response.dump(), "application/json");
    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "Pin/unpin failed: " << e.what() << std::endl;
        res.status = 400;
        nlohmann::json error = {{"error", {
            {"message", e.what()},
            {"type", "invalid_request_error"}
        }}};
        res.set_content(error.dump(), "application/json");
    }
}

void Server::handle_delete(const httplib::Request& req, httplib::Response& res) {
    nlohmann::json request_json;
    if (!parse_required_json_body(req, res, request_json)) return;

    try {
        // Accept both "model" and "model_name" for compatibility
        std::string model_name = request_json.contains("model") ?
            request_json["model"].get<std::string>() :
            request_json["model_name"].get<std::string>();

        LOG(INFO, "Server") << "Deleting model: " << model_name << std::endl;

        // If the model is currently loaded, unload it first to release file locks
        if (router_->is_model_loaded(model_name)) {
            LOG(INFO, "Server") << "Model is loaded, unloading before delete: " << model_name << std::endl;
            router_->unload_model(model_name);
        }

        // Retry delete with delays to handle in-progress downloads releasing file handles
        // This handles the race condition where a cancelled download hasn't yet released
        // its file handles when the delete request arrives
        const int max_retries = 3;
        const int retry_delay_seconds = 5;
        std::string last_error;

        for (int attempt = 0; attempt <= max_retries; ++attempt) {
            try {
                model_manager_->delete_model(model_name);

                // Success - send response and return
                nlohmann::json response = {
                    {"status", "success"},
                    {"message", "Deleted model: " + model_name}
                };
                res.set_content(response.dump(), "application/json");
                return;

            } catch (const std::exception& e) {
                last_error = e.what();

                // Only retry on "file in use" type errors (Windows and POSIX patterns)
                bool is_file_locked =
                    last_error.find("being used by another process") != std::string::npos ||
                    last_error.find("Permission denied") != std::string::npos ||
                    last_error.find("resource busy") != std::string::npos;

                if (is_file_locked && attempt < max_retries) {
                    LOG(INFO, "Server") << "Delete failed (file in use), retry "
                              << (attempt + 1) << "/" << max_retries
                              << " in " << retry_delay_seconds << "s..." << std::endl;
                    std::this_thread::sleep_for(std::chrono::seconds(retry_delay_seconds));
                    continue;
                }

                // Non-retryable error or max retries exceeded - rethrow
                throw;
            }
        }

    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "ERROR in handle_delete: " << e.what() << std::endl;

        // Check if this is a "Model not found" error (return 422)
        std::string error_msg = e.what();
        if (error_msg.find("Model not found") != std::string::npos ||
            error_msg.find("not supported") != std::string::npos) {
            res.status = 422;
        } else {
            res.status = 500;
        }

        nlohmann::json error = {{"error", e.what()}};
        res.set_content(error.dump(), "application/json");
    }
}

void Server::handle_cleanup_cache(const httplib::Request& req, httplib::Response& res) {
    try {
        auto request_json = nlohmann::json::parse(req.body);
        bool dry_run = request_json.value("dry_run", true);

        auto result = model_manager_->cleanup_orphaned_cache(dry_run);
        res.set_content(result.dump(), "application/json");
    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "ERROR in handle_cleanup_cache: " << e.what() << std::endl;
        res.status = 500;
        auto error_response = create_model_error("", e.what());
        res.set_content(error_response.dump(), "application/json");
    }
}

// Alias failures are not model-lookup failures. create_model_error() rewrites its
// message whenever the name is absent from the registry, which is always true for a
// new alias, so routing these through it would report "model not found" instead of
// the actual reason.
static nlohmann::json create_alias_error(const std::string& message, const std::string& code,
                                         const std::string& type = "invalid_request_error") {
    return nlohmann::json{{"error", {
        {"message", message},
        {"type", type},
        {"param", "alias"},
        {"code", code}
    }}};
}

void Server::handle_aliases_get(const httplib::Request& req, httplib::Response& res) {
    try {
        nlohmann::json alias_list = nlohmann::json::array();
        if (alias_manager_) {
            for (const auto& [alias, target] : alias_manager_->get_all_aliases()) {
                bool downloaded = false;
                std::string recipe = "llamacpp";
                try {
                    if (model_manager_->model_exists(target)) {
                        std::string canonical = model_manager_->resolve_model_name(target);
                        downloaded = model_manager_->is_model_downloaded(canonical);
                        ModelInfo info = model_manager_->get_model_info(canonical);
                        recipe = info.recipe;
                    }
                } catch (...) {}
                alias_list.push_back({
                    {"alias", alias},
                    {"target", target},
                    {"downloaded", downloaded},
                    {"recipe", recipe}
                });
            }
        }
        res.set_content(nlohmann::json{{"aliases", alias_list}}.dump(), "application/json");
    } catch (const std::exception& e) {
        res.status = 500;
        res.set_content(create_alias_error(e.what(), "internal_error", "server_error").dump(), "application/json");
    }
}

void Server::handle_aliases_add(const httplib::Request& req, httplib::Response& res) {
    try {
        auto req_json = nlohmann::json::parse(req.body);
        std::string alias = utils::normalize_model_name(req_json.value("alias", ""));
        std::string target = utils::normalize_model_name(req_json.value("target", req_json.value("model", "")));

        if (alias.empty() || target.empty()) {
            res.status = 400;
            res.set_content(create_alias_error("Alias and target fields are required", "invalid_request").dump(), "application/json");
            return;
        }

        if (alias == target) {
            res.status = 400;
            res.set_content(create_alias_error("Alias cannot point to itself", "invalid_alias").dump(), "application/json");
            return;
        }

        if (alias.rfind("user.", 0) == 0 || alias.rfind("extra.", 0) == 0 || alias.rfind("builtin.", 0) == 0) {
            res.status = 400;
            res.set_content(create_alias_error("Alias name cannot use reserved prefixes (user., extra., builtin.)", "invalid_alias").dump(), "application/json");
            return;
        }

        if (model_manager_->model_exists(alias)) {
            std::string canonical = model_manager_->resolve_model_name(alias);
            if (canonical == alias || canonical == "user." + alias || canonical == "builtin." + alias) {
                res.status = 409;
                res.set_content(create_alias_error("Cannot create alias '" + alias + "': Name conflicts with an existing canonical model", "alias_conflict").dump(), "application/json");
                return;
            }
        }

        if (!alias_manager_) {
            res.status = 500;
            res.set_content(create_alias_error("AliasManager uninitialized", "internal_error", "server_error").dump(), "application/json");
            return;
        }

        std::string err_msg;
        if (!alias_manager_->set_alias(alias, target, err_msg)) {
            res.status = 409;
            res.set_content(create_alias_error(err_msg, "alias_conflict").dump(), "application/json");
            return;
        }

        nlohmann::json response = {
            {"status", "ok"},
            {"alias", alias},
            {"target", target}
        };
        res.set_content(response.dump(), "application/json");
    } catch (const std::exception& e) {
        res.status = 400;
        res.set_content(create_alias_error(e.what(), "invalid_request").dump(), "application/json");
    }
}

void Server::handle_aliases_remove(const httplib::Request& req, httplib::Response& res) {
    std::string alias = utils::normalize_model_name(req.matches[1]);
    try {
        if (!alias_manager_ || !alias_manager_->remove_alias(alias)) {
            res.status = 404;
            res.set_content(create_alias_error("Alias not found: " + alias, "alias_not_found").dump(), "application/json");
            return;
        }
        nlohmann::json response = {
            {"status", "deleted"},
            {"alias", alias}
        };
        res.set_content(response.dump(), "application/json");
    } catch (const std::exception& e) {
        res.status = 500;
        res.set_content(create_alias_error(e.what(), "internal_error", "server_error").dump(), "application/json");
    }
}

std::string Server::resolve_alias_target(const std::string& model_name) const {
    if (alias_manager_) {
        auto resolved = alias_manager_->resolve_alias(model_name);
        if (resolved.has_value()) {
            return resolved.value();
        }
    }
    return model_name;
}

void Server::normalize_and_resolve_request_model(nlohmann::json& request_json) const {
    if (request_json.contains("model") && request_json["model"].is_string()) {
        std::string raw_name = request_json["model"].get<std::string>();
        std::string resolved = resolve_alias_target(raw_name);
        if (resolved != raw_name) {
            request_json["model"] = resolved;
        }
    }
    if (request_json.contains("model_name") && request_json["model_name"].is_string()) {
        std::string raw_name = request_json["model_name"].get<std::string>();
        std::string resolved = resolve_alias_target(raw_name);
        if (resolved != raw_name) {
            request_json["model_name"] = resolved;
        }
    }
}

void Server::persist_cloud_providers() {
    if (config_dir_.empty() || !cloud_registry_) return;
    try {
        json user_cfg = ConfigFile::load_raw(config_dir_);
        user_cfg["cloud_providers"] = cloud_registry_->to_config_array();
        ConfigFile::save(config_dir_, user_cfg);
    } catch (const std::exception& e) {
        // Persistence failure must not undo the in-memory change — the
        // registry is already updated and the provider is usable until
        // restart. Log so the operator can see why config.json is stale.
        LOG(WARNING, "Server") << "Failed to persist cloud_providers to config.json: "
                                << e.what() << std::endl;
    }
}

void Server::handle_cloud_auth_set(const httplib::Request& req, httplib::Response& res) {
    try {
        const auto body = nlohmann::json::parse(req.body);
        if (!body.contains("provider") || !body["provider"].is_string() ||
            !body.contains("api_key") || !body["api_key"].is_string()) {
            res.status = 400;
            nlohmann::json error = {{"error", {
                {"message", "Body must contain string fields: provider, api_key"},
                {"type", "invalid_request_error"}}}};
            res.set_content(error.dump(), "application/json");
            return;
        }
        const auto provider = body["provider"].get<std::string>();
        const auto api_key = body["api_key"].get<std::string>();
        bool allow_insecure_http = false;
        if (body.contains("allow_insecure_http")) {
            if (!body["allow_insecure_http"].is_boolean()) {
                res.status = 400;
                nlohmann::json error = {{"error", {
                    {"message", "allow_insecure_http must be a boolean when provided"},
                    {"type", "invalid_request_error"}}}};
                res.set_content(error.dump(), "application/json");
                return;
            }
            allow_insecure_http = body["allow_insecure_http"].get<bool>();
        }
        if (provider.empty() || api_key.empty()) {
            res.status = 400;
            nlohmann::json error = {{"error", {
                {"message", "provider and api_key must be non-empty"},
                {"type", "invalid_request_error"}}}};
            res.set_content(error.dump(), "application/json");
            return;
        }

        if (!cloud_registry_->is_installed(provider)) {
            res.status = 404;
            nlohmann::json error = {{"error", {
                {"message", "Cloud provider '" + provider + "' is not installed. "
                            "Call POST /v1/install with backend=cloud, provider, "
                            "and base_url first."},
                {"type", "invalid_request_error"}}}};
            res.set_content(error.dump(), "application/json");
            return;
        }

        const std::string base_url = cloud_registry_->base_url_for(provider);
        if (CloudProviderRegistry::is_http_base_url(base_url)) {
            const bool already_allowed = cloud_registry_->allow_insecure_http_for(provider);
            if (!allow_insecure_http && !already_allowed) {
                res.status = 400;
                nlohmann::json error = {{"error", {
                    {"message", "Cloud provider '" + provider + "' uses http://. "
                                "Set allow_insecure_http=true to explicitly opt in before "
                                "storing or using an API key over plaintext HTTP."},
                    {"type", "invalid_request_error"},
                    {"code", "insecure_http_requires_opt_in"}}}};
                res.set_content(error.dump(), "application/json");
                return;
            }
            if (allow_insecure_http && !already_allowed) {
                cloud_registry_->set_allow_insecure_http(provider, true);
                persist_cloud_providers();
            }
        }

        // env-wins-over-runtime: if the env var is set, refuse the runtime
        // key with 409 so the caller knows their POST had no effect.
        if (!cloud_registry_->set_runtime_key(provider, api_key)) {
            res.status = 409;
            const auto env_name = CloudProviderRegistry::env_var_name(provider);
            nlohmann::json error = {{"error", {
                {"message", env_name + " is set in the lemond process; the env var "
                            "takes precedence and the supplied API key was not stored."},
                {"type", "auth_conflict"},
                {"env_var", env_name}}}};
            res.set_content(error.dump(), "application/json");
            return;
        }

        // Refresh the provider's discovered model list with the new key.
        // Best-effort — refresh logs failures and returns 0. Either way the
        // key is stored and the response reflects the auth state.
        size_t models_after = model_manager_->refresh_cloud_models(provider);

        const auto state = cloud_registry_->auth_state(provider);
        nlohmann::json response = {
            {"provider", provider},
            {"allow_insecure_http", cloud_registry_->allow_insecure_http_for(provider)},
            {"auth_state", {
                {"env_var_set", state.env_var_set},
                {"runtime_key_set", state.runtime_key_set}
            }},
            {"models_discovered", models_after}
        };
        attach_warnings(
            response,
            CloudProviderRegistry::base_url_warnings(
                base_url,
                /*api_key_available=*/true));
        res.set_content(response.dump(), "application/json");
    } catch (const nlohmann::json::parse_error& e) {
        res.status = 400;
        nlohmann::json error = {{"error", {{"message", "Invalid JSON: " + std::string(e.what())},
                                            {"type", "invalid_request_error"}}}};
        res.set_content(error.dump(), "application/json");
    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "ERROR in handle_cloud_auth_set: " << e.what() << std::endl;
        res.status = 500;
        nlohmann::json error = {{"error", {{"message", e.what()}, {"type", "server_error"}}}};
        res.set_content(error.dump(), "application/json");
    }
}

void Server::handle_cloud_auth_clear(const httplib::Request& req, httplib::Response& res) {
    try {
        const std::string provider = req.matches[1].str();
        if (provider.empty()) {
            res.status = 400;
            nlohmann::json error = {{"error", {{"message", "Missing provider in URL"},
                                                {"type", "invalid_request_error"}}}};
            res.set_content(error.dump(), "application/json");
            return;
        }

        const bool cleared = cloud_registry_->clear_runtime_key(provider);
        // If the env var is set, models stay discovered (env still authenticates);
        // if not, the cache should reflect the now-unauthenticated state.
        const auto state = cloud_registry_->auth_state(provider);
        if (!state.env_var_set) {
            model_manager_->evict_cloud_models(provider);
        }

        nlohmann::json response = {
            {"provider", provider},
            {"cleared_runtime_key", cleared},
            {"auth_state", {
                {"env_var_set", state.env_var_set},
                {"runtime_key_set", state.runtime_key_set}
            }}
        };
        res.set_content(response.dump(), "application/json");
    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "ERROR in handle_cloud_auth_clear: " << e.what() << std::endl;
        res.status = 500;
        nlohmann::json error = {{"error", {{"message", e.what()}, {"type", "server_error"}}}};
        res.set_content(error.dump(), "application/json");
    }
}

// Called by handle_pull when local_import=true
// Parameters:
//   - dest_path: Directory where model files are located (already copied/uploaded)
//   - model_name: Model name with "user." prefix
//   - model_data: Request content
//   - hf_cache: HuggingFace cache directory for computing relative paths
void Server::resolve_and_register_local_model(
    const std::string& dest_path,
    const std::string& model_name,
    const json& model_data,
    const std::string& hf_cache) {
    std::string mmproj = model_data.value("mmproj", "");
    std::string recipe = model_data.value("recipe", "");
    bool vision = model_data.value("vision", false);

    // The backend's ops locate its primary artifact within the imported
    // directory (.gguf / .bin file, genai_config.json dir, …); "" means register
    // the directory itself.
    std::string resolved_checkpoint = backends::ops_for(recipe)->find_imported_checkpoint(dest_path);
    std::string resolved_mmproj;

    // Search for mmproj file if vision is enabled or mmproj hint provided
    if (vision || !mmproj.empty()) {
        for (const auto& entry : std::filesystem::recursive_directory_iterator(dest_path)) {
            if (entry.is_regular_file()) {
                std::string filename = entry.path().filename().string();
                std::string filename_lower = filename;
                std::transform(filename_lower.begin(), filename_lower.end(), filename_lower.begin(), ::tolower);

                // Match either the provided mmproj name or any mmproj file
                if (!mmproj.empty() && filename == mmproj) {
                    resolved_mmproj = filename;
                    vision = true;  // Ensure vision is set
                    break;
                } else if (filename_lower.find("mmproj") != std::string::npos) {
                    resolved_mmproj = filename;
                    vision = true;  // Ensure vision is set
                    break;
                }
            }
        }
    }

    // Build checkpoint for registration - store as relative path from HF cache
    std::string checkpoint_to_register;
    std::filesystem::path hf_cache_path = utils::path_from_utf8(hf_cache);
    if (!resolved_checkpoint.empty()) {
        std::filesystem::path rel = std::filesystem::relative(
            utils::path_from_utf8(resolved_checkpoint), hf_cache_path);
        checkpoint_to_register = utils::path_to_utf8(rel);
    } else {
        // Fallback - use dest_path relative to hf_cache
        std::filesystem::path rel = std::filesystem::relative(
            utils::path_from_utf8(dest_path), hf_cache_path);
        checkpoint_to_register = utils::path_to_utf8(rel);
    }

    LOG(INFO, "Server") << "Registering model with checkpoint: " << checkpoint_to_register << std::endl;

    auto actual_model_data = model_data;
    actual_model_data["checkpoint"] = checkpoint_to_register;
    if (!resolved_mmproj.empty()) {
        actual_model_data["mmproj"] = resolved_mmproj;
    }

    // Register the model with source to mark how it was added
    model_manager_->register_user_model(
        model_name,
        actual_model_data,
        "local_upload"
    );

    LOG(INFO, "Server") << "Model registered successfully" << std::endl;
}

void Server::handle_stats(const httplib::Request& req, httplib::Response& res) {
    // For HEAD requests, just return 200 OK without processing
    if (req.method == "HEAD") {
        res.status = 200;
        return;
    }

    try {
        auto stats = router_->get_stats();
        res.set_content(stats.dump(), "application/json");
    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "ERROR in handle_stats: " << e.what() << std::endl;
        res.status = 500;
        nlohmann::json error = {{"error", e.what()}};
        res.set_content(error.dump(), "application/json");
    }
}

void Server::handle_metrics(const httplib::Request& req, httplib::Response& res) {
    if (req.method == "HEAD") {
        res.status = 200;
        return;
    }

    try {
        SystemMetrics system_metrics;
        system_metrics.cpu_percent = get_cpu_usage();
        system_metrics.gpu_percent = get_gpu_usage();
        system_metrics.vram_gb = get_vram_usage();
        system_metrics.npu_percent = get_npu_utilization();

        res.set_content(build_prometheus_metrics(*router_, system_metrics),
                        "text/plain; version=0.0.4; charset=utf-8");
    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "ERROR in handle_metrics: " << e.what() << std::endl;
        res.status = 500;
        res.set_content("# Lemonade metrics error\n", "text/plain; version=0.0.4; charset=utf-8");
    }
}

void Server::handle_system_info(const httplib::Request& req, httplib::Response& res) {
    // For HEAD requests, just return 200 OK without processing
    if (req.method == "HEAD") {
        res.status = 200;
        return;
    }

    // SystemInfoCache is the single source of truth for hardware + recipes.
    // Recipes are cached until invalidated by install/uninstall.
    nlohmann::json system_info = SystemInfoCache::get_system_info_with_cache();

    // Enrich with release_url, download_filename, version from BackendManager config
    if (system_info.contains("recipes")) {
        enrich_recipes(system_info["recipes"]);
    }

    // Surfaced as a separate host-specific field rather than by pruning
    // `recipes`, which must stay canonical: the docs generator renders
    // README/models.js from it and would otherwise drift with the generating
    // machine's memory. Dynamic-model backends (cloud, flm) are never listed.
    if (model_manager_) {
        nlohmann::json unavailable = nlohmann::json::array();
        for (const std::string& recipe : model_manager_->recipes_with_all_models_filtered()) {
            const BackendDescriptor* desc = backends::descriptor_for(recipe);
            if (desc && desc->dynamic_models) {
                continue;
            }
            unavailable.push_back(recipe);
        }
        system_info["unavailable_recipes"] = std::move(unavailable);
    }

    // Surface runtime config flags that affect client-side install/download UX.
    if (auto* cfg = RuntimeConfig::global()) {
        system_info["no_fetch_executables"] = cfg->no_fetch_executables();
    }

    if (config_) {
        system_info["model_storage"] = get_model_storage_stats(utils::get_hf_cache_dir());
    }

    // Cloud providers: per-provider {name, base_url, env_var_set,
    // runtime_key_set, models_discovered}. Never includes the key itself.
    // Clients use this to decide whether to prompt for an API key, show a
    // "configured by env var" badge, or surface the model count to the user.
    if (cloud_registry_) {
        nlohmann::json providers = nlohmann::json::array();
        for (const auto& rec : cloud_registry_->list_installed()) {
            auto state = cloud_registry_->auth_state(rec.name);
            nlohmann::json provider = {
                {"name", rec.name},
                {"base_url", rec.base_url},
                {"allow_insecure_http", rec.allow_insecure_http},
                {"auth_header_name", rec.auth_header_name},
                {"auth_header_prefix", rec.auth_header_prefix},
                {"wire_format", rec.wire_format},
                {"env_var", CloudProviderRegistry::env_var_name(rec.name)},
                {"env_var_set", state.env_var_set},
                {"runtime_key_set", state.runtime_key_set},
                {"models_discovered", model_manager_->count_cloud_models(rec.name)}
            };
            attach_warnings(
                provider,
                CloudProviderRegistry::base_url_warnings(
                    rec.base_url,
                    state.env_var_set || state.runtime_key_set));
            providers.push_back(std::move(provider));
        }
        system_info["cloud"] = {{"providers", providers}};
    }

    res.set_content(system_info.dump(), "application/json");
}

// Get CPU usage percentage
double Server::get_cpu_usage() {
#if defined(__linux__) || defined(_WIN32)
    return metrics_platform_->get_cpu_usage(cpu_stats_mutex_,
                                            last_cpu_stats_.total,
                                            last_cpu_stats_.total_idle);
#else
    uint64_t dummy_total = 0, dummy_idle = 0;
    std::mutex dummy_mutex;
    return metrics_platform_->get_cpu_usage(dummy_mutex, dummy_total, dummy_idle);
#endif
}

// Get GPU usage percentage (AMD GPUs on Linux)
double Server::get_gpu_usage() {
    return metrics_platform_->get_gpu_usage();
}

// Get VRAM/GTT usage in GB (AMD GPUs on Linux)
double Server::get_vram_usage() {
    return metrics_platform_->get_vram_usage_gb();
}

// Helper: Get NPU utilization (AMD NPU on Linux)
double Server::get_npu_utilization() {
    return metrics_platform_->get_npu_utilization();
}

void Server::handle_system_stats(const httplib::Request& req, httplib::Response& res) {
    // For HEAD requests, just return 200 OK without processing
    if (req.method == "HEAD") {
        res.status = 200;
        return;
    }

    nlohmann::json stats;

    // CPU usage
    double cpu_percent = get_cpu_usage();
    stats["cpu_percent"] = (cpu_percent >= 0) ? nlohmann::json(cpu_percent) : nlohmann::json();

    // Get memory info
    stats["memory_gb"] = metrics_platform_->get_memory_usage_gb();

    // GPU usage
    double gpu_percent = get_gpu_usage();
    stats["gpu_percent"] = (gpu_percent >= 0) ? nlohmann::json(gpu_percent) : nlohmann::json();

    // VRAM usage
    double vram_gb = get_vram_usage();
    stats["vram_gb"] = (vram_gb >= 0) ? nlohmann::json(vram_gb) : nlohmann::json();

    // NPU Utilization
    double npu_percent = get_npu_utilization();
    stats["npu_percent"] = (npu_percent >= 0) ? nlohmann::json(npu_percent) : nlohmann::json();

    res.set_content(stats.dump(), "application/json");
}

void Server::handle_simulate_vram_pressure(const httplib::Request& req, httplib::Response& res) {
    try {
        auto req_json = json::parse(req.body);
        double pct = req_json.value("pct", 0.0);
        router_->simulate_vram_pressure(pct);
        res.set_content(R"({"status": "ok"})", "application/json");
    } catch (const std::exception& e) {
        res.status = 400;
        res.set_content(json{{"error", e.what()}}.dump(), "application/json");
    }
}

namespace {

std::string job_id_from_path(const httplib::Request& req) {
    return req.matches.size() > 1 ? std::string(req.matches[1]) : std::string();
}

void job_error(httplib::Response& res, int status, const std::string& message) {
    res.status = status;
    res.set_content(lemon::jobs::json{{"error", message}}.dump(), "application/json");
}

}

void Server::handle_jobs_create(const httplib::Request& req, httplib::Response& res) {
    try {
        auto body = lemon::jobs::json::parse(req.body);
        const std::string name = body.value("name", "");
        lemon::jobs::json steps_json = lemon::jobs::json::array();
        if (body.contains("definition") && body["definition"].is_object()
            && body["definition"].contains("steps")) {
            steps_json = body["definition"]["steps"];
        } else if (body.contains("steps")) {
            steps_json = body["steps"];
        }
        std::vector<lemon::jobs::StepRecord> steps;
        if (steps_json.is_array())
            for (const auto& s : steps_json)
                steps.push_back(lemon::jobs::StepRecord::from_json(s));
        lemon::jobs::json inputs =
            body.contains("inputs") ? body["inputs"] : lemon::jobs::json::object();
        const std::string id = job_manager_->create(name, std::move(steps), inputs);
        res.status = 202;
        res.set_content(lemon::jobs::json{{"id", id}}.dump(), "application/json");
    } catch (const lemon::jobs::JobError& e) {
        job_error(res, e.status, e.what());
    } catch (const std::exception& e) {
        job_error(res, 400, e.what());
    }
}

void Server::handle_jobs_list(const httplib::Request&, httplib::Response& res) {
    res.set_content(lemon::jobs::json{{"jobs", job_manager_->list()}}.dump(), "application/json");
}

void Server::handle_jobs_get(const httplib::Request& req, httplib::Response& res) {
    const auto job = job_manager_->get(job_id_from_path(req));
    if (!job) {
        job_error(res, 404, "unknown job");
        return;
    }
    res.set_content(job->dump(), "application/json");
}

void Server::handle_jobs_pause(const httplib::Request& req, httplib::Response& res) {
    if (!job_manager_->pause(job_id_from_path(req))) {
        job_error(res, 404, "job not found or not pausable");
        return;
    }
    res.set_content(R"({"status":"pausing"})", "application/json");
}

void Server::handle_jobs_interrupt(const httplib::Request& req, httplib::Response& res) {
    if (!job_manager_->interrupt(job_id_from_path(req))) {
        job_error(res, 404, "job not found or not interruptible");
        return;
    }
    res.set_content(R"({"status":"interrupting"})", "application/json");
}

void Server::handle_jobs_resume(const httplib::Request& req, httplib::Response& res) {
    if (!job_manager_->resume(job_id_from_path(req))) {
        job_error(res, 404, "job not found or not resumable");
        return;
    }
    res.set_content(R"({"status":"resuming"})", "application/json");
}

void Server::handle_jobs_delete(const httplib::Request& req, httplib::Response& res) {
    bool active = false;
    if (!job_manager_->remove(job_id_from_path(req), active)) {
        job_error(res, active ? 409 : 404, active ? "job is active" : "unknown job");
        return;
    }
    res.set_content(R"({"status":"deleted"})", "application/json");
}

void Server::handle_shutdown(const httplib::Request& req, httplib::Response& res) {
    LOG(INFO, "Server") << "Shutdown request received" << std::endl;

    // Unload all models SYNCHRONOUSLY before sending the response.
    // This ensures child processes (llama-server, etc.) are terminated
    // before the caller proceeds, avoiding zombie processes.
    if (router_) {
        LOG(INFO, "Server") << "Unloading models and stopping backend servers..." << std::endl;
        try {
            router_->unload_model();
            LOG(INFO, "Server") << "All models unloaded" << std::endl;
        } catch (const std::exception& e) {
            LOG(ERROR, "Server") << "Error during unload: " << e.what() << std::endl;
        }
    }

    nlohmann::json response = {{"status", "shutting down"}};
    res.set_content(response.dump(), "application/json");

    // Stop the HTTP listener and exit asynchronously (allows response to be sent first)
    std::thread([this]() {
        std::this_thread::sleep_for(std::chrono::milliseconds(100));
        cancel_download_jobs();
        stop();
        std::exit(0);
    }).detach();
}

void Server::handle_config_set(const httplib::Request& req, httplib::Response& res) {
    try {
        auto body = nlohmann::json::parse(req.body);

        auto result = config_->set(body, [this](const json& applied) {
            apply_config_side_effects(applied);
        });

        if (!config_dir_.empty()) {
            try {
                json user_cfg = ConfigFile::load_raw(config_dir_);
                if (result.contains("updated") && result["updated"].is_object()) {
                    user_cfg = utils::JsonUtils::merge(user_cfg, result["updated"]);
                }
                json defaults = ConfigFile::get_defaults();
                utils::JsonUtils::prune_matching(user_cfg, defaults);
                ConfigFile::save(config_dir_, user_cfg);
            } catch (const std::exception& e) {
                LOG(WARNING, "Server") << "Failed to persist config.json: " << e.what() << std::endl;
            }
        }

        res.set_content(result.dump(), "application/json");
    } catch (const nlohmann::json::parse_error& e) {
        res.status = 400;
        nlohmann::json error = {{"error", "Invalid JSON in request body"}};
        res.set_content(error.dump(), "application/json");
    } catch (const std::invalid_argument& e) {
        res.status = 400;
        nlohmann::json error = {{"error", e.what()}};
        res.set_content(error.dump(), "application/json");
    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "ERROR in handle_config_set: " << e.what() << std::endl;
        res.status = 500;
        nlohmann::json error = {{"error", e.what()}};
        res.set_content(error.dump(), "application/json");
    }
}

void Server::handle_config_get(const httplib::Request& /*req*/, httplib::Response& res) {
    try {
        auto snap = config_->snapshot();
        res.set_content(snap.dump(), "application/json");
    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "ERROR in handle_config_get: " << e.what() << std::endl;
        res.status = 500;
        nlohmann::json error = {{"error", e.what()}};
        res.set_content(error.dump(), "application/json");
    }
}

void Server::handle_config_defaults_get(const httplib::Request& /*req*/, httplib::Response& res) {
    try {
        // The canonical default config (global keys + descriptor-derived per-recipe
        // sections), independent of this host's config.json or deployment overrides.
        res.set_content(ConfigFile::base_defaults().dump(2), "application/json");
    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "ERROR in handle_config_defaults_get: " << e.what() << std::endl;
        res.status = 500;
        nlohmann::json error = {{"error", e.what()}};
        res.set_content(error.dump(), "application/json");
    }
}

void Server::handle_bin_change(const std::string& section,
                                const std::string& bin_key,
                                const std::string& new_value) {
    std::string recipe = RuntimeConfig::config_section_to_recipe(section);

    // bin_key is "<backend>_bin" — strip the suffix to get the backend name
    // expected by install_backend / find_external_backend_binary.
    std::string backend = bin_key.substr(0, bin_key.size() - 4);

    // The "server_bin" key (as in ryzenai.server_bin) is not consumed by the
    // current install flow, so skip the hot-swap rather than attempt an install
    // that won't help.
    if (backend == "server") {
        LOG(WARNING, "Server") << section << "." << bin_key
                               << " is not consumed by the install flow; "
                                  "no hot-swap performed." << std::endl;
        return;
    }

    LOG(INFO, "Server") << "*_bin config changed: " << section << "." << bin_key
                        << " = '" << new_value << "' — hot-swapping "
                        << recipe << ":" << backend << std::endl;

    // Snapshot loaded models on this (recipe, backend). A model whose options
    // do not pin a backend (mb empty) is treated as potentially affected since
    // it could resolve to the changed backend on next load.
    struct Saved {
        std::string name;
        RecipeOptions opts;
    };
    std::vector<Saved> previously_loaded;
    auto loaded = router_->get_all_loaded_models();
    std::string backend_option_key = recipe + "_backend";
    for (const auto& m : loaded) {
        if (m.value("recipe", "") != recipe) continue;
        std::string mb;
        if (m.contains("recipe_options") && m["recipe_options"].contains(backend_option_key)) {
            mb = m["recipe_options"].value(backend_option_key, "");
        }
        if (!mb.empty() && mb != backend) continue;
        std::string name = m.value("model_name", "");
        if (name.empty()) continue;
        previously_loaded.push_back({name, router_->get_model_recipe_options(name)});
    }

    for (const auto& s : previously_loaded) {
        LOG(INFO, "Server") << "Unloading " << s.name
                            << " before installing new " << recipe << ":" << backend
                            << " binary" << std::endl;
        try {
            router_->unload_model(s.name);
        } catch (const std::exception& e) {
            LOG(WARNING, "Server") << "Failed to unload " << s.name << ": " << e.what() << std::endl;
        }
    }

    // Install the new binary. install_from_github bails early when the user's
    // value resolves to a path (find_external_backend_binary returns it). When
    // version.txt mismatches the resolved version, the install dir is wiped
    // and re-downloaded.
    try {
        backend_manager_->install_backend(recipe, backend);
    } catch (const std::exception& e) {
        LOG(WARNING, "Server") << "install_backend(" << recipe << ":" << backend
                               << ") failed after *_bin change: " << e.what() << std::endl;
    }

    // Best-effort reload of previously-loaded models on the new binary.
    for (const auto& s : previously_loaded) {
        try {
            auto info = model_manager_->get_model_info(s.name);
            router_->load_model(s.name, info, s.opts, true);
            LOG(INFO, "Server") << "Reloaded " << s.name << " on new "
                                << recipe << ":" << backend << " binary" << std::endl;
        } catch (const std::exception& e) {
            LOG(WARNING, "Server") << "Failed to reload " << s.name
                                   << " after *_bin hot-swap: " << e.what() << std::endl;
        }
    }

    SystemInfoCache::invalidate_recipes();
    model_manager_->invalidate_models_cache();
}

void Server::apply_config_side_effects(const json& applied_changes) {
    for (auto& [key, value] : applied_changes.items()) {
        if (key == "port") {
            int new_port = config_->port();
            int current_port = port_.load();
            if (new_port != current_port) {
                LOG(INFO, "Server") << "Port change requested: " << current_port << " -> " << new_port << std::endl;
                port_.store(new_port);
                if (running_) {
                    rebind_requested_ = true;
                    udp_beacon_.stopBroadcasting();
                    stop_http_listeners();
                }
            }
        } else if (key == "host") {
            LOG(INFO, "Server") << "Host change requested to: " << config_->host() << std::endl;
            if (running_) {
                rebind_requested_ = true;
                udp_beacon_.stopBroadcasting();
                stop_http_listeners();
            }
            // Restart websocket server with new host
            if (websocket_server_ && running_) {
                websocket_server_->stop();
                websocket_server_ = std::make_unique<WebSocketServer>(
                    router_.get(),
                    config_->host(),
                    config_->websocket_port());
                websocket_server_->start();
            }
        } else if (key == "websocket_port") {
            if (websocket_server_) {
                LOG(INFO, "Server") << "Restarting WebSocket server on requested port "
                                    << config_->websocket_port() << std::endl;
                websocket_server_->stop();
                websocket_server_ = std::make_unique<WebSocketServer>(
                    router_.get(),
                    config_->host(),
                    config_->websocket_port());
                if (running_) {
                    websocket_server_->start();
                }
            }
        } else if (key == "log_level") {
            std::string level = config_->log_level();
            LOG(INFO, "Server") << "Log level changed to: " << level << std::endl;
            reconfigure_application_logging(level);
        } else if (key == "global_timeout") {
            long timeout = config_->global_timeout();
            LOG(INFO, "Server") << "Global timeout changed to: " << timeout << "s" << std::endl;
            utils::HttpClient::set_default_timeout(timeout);
        } else if (key == "download_rate_limit") {
            const int64_t bps = config_->download_rate_limit_bytes_per_second();
            if (bps > 0) {
                LOG(INFO, "Server") << "Download rate limit enabled at " << bps << " B/s" << std::endl;
            } else {
                LOG(INFO, "Server") << "Download rate limit disabled" << std::endl;
            }
            utils::HttpClient::set_download_rate_limit(bps);
        } else if (key == "broadcast" || key == "no_broadcast") {
            bool bcast = config_->broadcast();
            LOG(INFO, "Server") << "Broadcast " << (bcast ? "enabled" : "disabled") << std::endl;
            if (!bcast) {
                udp_beacon_.stopBroadcasting();
            } else {
                auto rfc1918Interfaces = udp_beacon_.getLocalRFC1918Interfaces();
                if (!rfc1918Interfaces.empty()) {
                    udp_beacon_.startBroadcasting(13305, port_, 2);
                }
            }
        } else if (key == "extra_models_dir") {
            std::string dir = config_->extra_models_dir();
            LOG(INFO, "Server") << "Extra models dir changed to: " << dir << std::endl;
            model_manager_->set_extra_models_dir(dir);
        } else if (key == "models_dir") {
            std::string dir = config_->models_dir();
            LOG(INFO, "Server") << "Models dir changed to: " << dir << std::endl;
            utils::set_models_dir(dir);
            model_manager_->invalidate_models_cache();
        } else if (key == "telemetry") {
            if (value.is_object()) {
                if (value.contains("enabled")) {
                    bool enabled = config_->telemetry_enabled();
                    LOG(INFO, "Server") << "Telemetry " << (enabled ? "enabled" : "disabled") << std::endl;
                }
                if (value.contains("otlp") && value["otlp"].is_object() && value["otlp"].contains("endpoint")) {
                    LOG(INFO, "Server") << "Telemetry endpoint changed to: " << config_->telemetry_otlp_endpoint() << std::endl;
                }
            }
        } else if (value.is_object()) {
            // Nested backend section change (llamacpp / whispercpp / sdcpp / ryzenai / kokoro).
            // Recipe defaults (e.g. default_backend) are derived from these settings, so
            // drop the memoized recipes so the next /system-info recomputes them.
            SystemInfoCache::invalidate_recipes();
            // Look for *_bin sub-keys and trigger a hot-swap of the affected backend.
            for (auto& [sub_key, sub_value] : value.items()) {
                if (sub_key.size() >= 4
                    && sub_key.compare(sub_key.size() - 4, 4, "_bin") == 0) {
                    handle_bin_change(key, sub_key,
                                      sub_value.is_string() ? sub_value.get<std::string>() : "");
                }
            }
        }
    }
}


// ============================================================================
// Server-owned Download Manager state
// ============================================================================

namespace {

constexpr auto DOWNLOAD_TERMINAL_VISIBILITY = std::chrono::seconds(30);

} // namespace

nlohmann::json Server::download_progress_to_json(const DownloadProgress& p) {
    nlohmann::json event_data;
    event_data["file"] = p.file;
    event_data["file_index"] = p.file_index;
    event_data["total_files"] = p.total_files;
    event_data["bytes_downloaded"] = static_cast<uint64_t>(p.bytes_downloaded);
    event_data["bytes_total"] = static_cast<uint64_t>(p.bytes_total);
    event_data["total_download_size"] = static_cast<uint64_t>(p.total_download_size);
    event_data["bytes_previously_downloaded"] = static_cast<uint64_t>(p.bytes_previously_downloaded);
    event_data["percent"] = p.percent;
    event_data["complete"] = p.complete;
    if (!p.error.empty()) {
        event_data["error"] = p.error;
    }
    return event_data;
}

nlohmann::json Server::download_job_to_json(const std::shared_ptr<DownloadJob>& job) {
    nlohmann::json item = job->progress.is_object() ? job->progress : nlohmann::json::object();
    item["id"] = job->id;
    item["type"] = job->type;
    item["model_name"] = job->display_name;
    item["status"] = job->status;
    item["running"] = job->running;
    if (!job->error.empty()) {
        item["error"] = job->error;
    }
    return item;
}

bool Server::is_download_job_visible(const std::shared_ptr<DownloadJob>& job) const {
    if (!job) return false;
    if (job->running) {
        return true;
    }
    if (job->status == "downloading" || job->status == "paused" || job->status == "error") {
        return true;
    }
    if (job->status == "cancelled") {
        // Cancelled downloads may still have partial files on disk. Keep them
        // discoverable across reloads/restarts until the UI explicitly removes
        // the row after cleanup, retry, or user dismissal.
        return true;
    }
    if (job->status == "completed") {
        return job->terminal_since.time_since_epoch().count() > 0 &&
               std::chrono::steady_clock::now() - job->terminal_since < DOWNLOAD_TERMINAL_VISIBILITY;
    }
    return false;
}

void Server::join_download_job(const std::shared_ptr<DownloadJob>& job) {
    // Download workers capture `this`, so they must never outlive Server. Move
    // the thread out under the job-local mutex, then join without holding either
    // worker_mutex or downloads_mutex_. This avoids both data races on
    // std::thread and deadlocks with progress callbacks.
    if (!job) return;

    std::thread worker;
    {
        std::lock_guard<std::mutex> lock(job->worker_mutex);
        if (!job->worker.joinable()) return;
        if (job->worker.get_id() == std::this_thread::get_id()) return;
        worker = std::move(job->worker);
    }

    worker.join();
}

std::shared_ptr<Server::DownloadJob> Server::start_download_job(
    const std::string& download_id,
    const std::string& download_type,
    const std::string& display_name,
    std::function<void(DownloadProgressCallback)> operation) {

    std::shared_ptr<DownloadJob> old_job;
    auto job = std::make_shared<DownloadJob>();
    job->id = download_id;
    job->type = download_type;
    job->display_name = display_name;
    job->status = "downloading";
    job->running = true;
    job->progress = {
        {"id", download_id},
        {"type", download_type},
        {"model_name", display_name},
        {"file", ""},
        {"file_index", 0},
        {"total_files", 0},
        {"bytes_downloaded", 0},
        {"bytes_total", 0},
        {"total_download_size", 0},
        {"bytes_previously_downloaded", 0},
        {"completed_files_bytes", 0},
        {"cumulative_bytes_downloaded", 0},
        {"overall_bytes_downloaded", 0},
        {"percent", 0},
        {"complete", false}
    };

    std::unique_lock<std::mutex> worker_lock(job->worker_mutex);

    {
        std::lock_guard<std::mutex> lock(downloads_mutex_);
        auto existing = download_jobs_.find(download_id);
        if (existing != download_jobs_.end()) {
            if (existing->second->running || existing->second->status == "downloading") {
                return existing->second;
            }
            old_job = existing->second;
        }
        download_jobs_[download_id] = job;
    }

    join_download_job(old_job);

    job->worker = std::thread([this, job, operation = std::move(operation)]() mutable {
        try {
            DownloadProgressCallback progress_cb = [this, job](const DownloadProgress& p) -> bool {
                std::lock_guard<std::mutex> lock(downloads_mutex_);
                // Completion wins over a very late pause/cancel request. If the
                // downloader is reporting its final complete event, record it and
                // let the operation return successfully instead of turning a fully
                // written model into a cancelled/paused row.
                if (job->cancel_requested && !p.complete) {
                    job->stop_acknowledged = true;
                    return false;
                }

                if (job->current_file_index != p.file_index) {
                    if (job->current_file_index >= 0 && p.file_index > job->current_file_index) {
                        job->completed_files_bytes += job->current_file_bytes_total;
                    }
                    job->current_file_index = p.file_index;
                    job->current_file_bytes_total = 0;
                }
                if (p.bytes_total > 0) {
                    job->current_file_bytes_total = std::max<uint64_t>(
                        job->current_file_bytes_total,
                        static_cast<uint64_t>(p.bytes_total));
                }

                const uint64_t total_download_size = static_cast<uint64_t>(p.total_download_size);
                uint64_t cumulative_bytes = job->completed_files_bytes + static_cast<uint64_t>(p.bytes_downloaded);
                if (p.complete && total_download_size > 0) {
                    cumulative_bytes = total_download_size;
                } else if (total_download_size > 0) {
                    cumulative_bytes = std::min(cumulative_bytes, total_download_size);
                }

                job->progress = download_progress_to_json(p);
                job->progress["completed_files_bytes"] = job->completed_files_bytes;
                job->progress["cumulative_bytes_downloaded"] = cumulative_bytes;
                job->progress["overall_bytes_downloaded"] = cumulative_bytes;
                job->status = p.complete ? "completed" : "downloading";
                // terminal_since is the time the worker has actually stopped, not
                // the time a terminal status first becomes visible. Keeping it
                // empty while running=true prevents /downloads from expiring the
                // row before other tabs can observe that files are released.
                job->terminal_since = std::chrono::steady_clock::time_point{};
                job->error.clear();
                return true;
            };

            operation(progress_cb);

            std::lock_guard<std::mutex> lock(downloads_mutex_);
            if (job->cancel_requested && job->stop_acknowledged) {
                job->status = job->cancel_action == "cancel" ? "cancelled" : "paused";
                job->progress["complete"] = false;
            } else {
                // The operation returned normally without acknowledging a stop
                // request. Treat that as success; a late pause/cancel button must
                // not erase or delete a completed download.
                job->status = "completed";
                job->progress["complete"] = true;
                job->progress["percent"] = 100;
                const uint64_t total_download_size = job->progress.value("total_download_size", uint64_t{0});
                if (total_download_size > 0) {
                    job->progress["cumulative_bytes_downloaded"] = total_download_size;
                    job->progress["overall_bytes_downloaded"] = total_download_size;
                }
            }
            job->running = false;
            job->terminal_since = (job->status == "completed" || job->status == "cancelled")
                ? std::chrono::steady_clock::now()
                : std::chrono::steady_clock::time_point{};
        } catch (const lemon::UnknownModelError& e) {
            std::lock_guard<std::mutex> lock(downloads_mutex_);
            LOG(ERROR, "DownloadManager") << "worker unknown-model error id=" << job->id
                                           << " error=\"" << e.what() << "\"" << std::endl;
            job->status = "error";
            job->terminal_since = std::chrono::steady_clock::time_point{};
            job->error = e.what();
            job->progress["error"] = e.what();
            job->progress["code"] = lemon::kUnknownModelErrorCode;
            job->running = false;
        } catch (const std::exception& e) {
            bool cancel_requested = false;
            {
                std::lock_guard<std::mutex> lock(downloads_mutex_);
                cancel_requested = job->cancel_requested;
            }

            if (!cancel_requested) {
                LOG(ERROR, "DownloadManager") << "worker exception id=" << job->id
                                               << " error=\"" << e.what() << "\"" << std::endl;
            }

            std::lock_guard<std::mutex> lock(downloads_mutex_);
            if (job->cancel_requested) {
                job->status = job->cancel_action == "cancel" ? "cancelled" : "paused";
                job->error.clear();
            } else {
                job->status = "error";
                job->terminal_since = std::chrono::steady_clock::time_point{};
                job->error = e.what();
                job->progress["error"] = e.what();
            }
            job->running = false;
            if (job->status == "cancelled") {
                job->terminal_since = std::chrono::steady_clock::now();
            }
        }
    });
    worker_lock.unlock();

    return job;
}


// ============================================================================
// Shared SSE streaming helper for download operations
// ============================================================================

void Server::stream_download_operation(
    httplib::Response& res,
    std::function<void(DownloadProgressCallback)> operation) {

    res.set_header("Cache-Control", "no-cache");
    res.set_header("Connection", "keep-alive");
    res.set_header("X-Accel-Buffering", "no");

    res.set_chunked_content_provider(
        "text/event-stream",
        [operation = std::move(operation)](size_t offset, httplib::DataSink& sink) {
            if (offset > 0) {
                return false; // Already sent everything
            }

            try {
                // Create progress callback that emits SSE events
                bool complete_sent = false;
                DownloadProgressCallback progress_cb = [&sink, &complete_sent](const DownloadProgress& p) -> bool {
                    nlohmann::json event_data;
                    event_data["file"] = p.file;
                    event_data["file_index"] = p.file_index;
                    event_data["total_files"] = p.total_files;
                    event_data["bytes_downloaded"] = static_cast<uint64_t>(p.bytes_downloaded);
                    event_data["bytes_total"] = static_cast<uint64_t>(p.bytes_total);
                    event_data["total_download_size"] = static_cast<uint64_t>(p.total_download_size);
                    event_data["bytes_previously_downloaded"] = static_cast<uint64_t>(p.bytes_previously_downloaded);
                    event_data["percent"] = p.percent;

                    std::string event;
                    if (p.complete) {
                        event = "event: complete\ndata: " + event_data.dump() + "\n\n";
                        complete_sent = true;
                    } else {
                        event = "event: progress\ndata: " + event_data.dump() + "\n\n";
                    }

                    if (!sink.write(event.c_str(), event.size())) {
                        LOG(INFO, "Server") << "Client disconnected, cancelling download" << std::endl;
                        return false;
                    }
                    return true;
                };

                operation(progress_cb);

                // If operation completed without sending a "complete" event
                // (e.g. backend was already installed), send one now
                if (!complete_sent) {
                    nlohmann::json done_data = {{"status", "ok"}};
                    std::string event = "event: complete\ndata: " + done_data.dump() + "\n\n";
                    sink.write(event.c_str(), event.size());
                }

            } catch (const lemon::UnknownModelError& e) {
                nlohmann::json error_data = {{"error", e.what()}, {"code", lemon::kUnknownModelErrorCode}};
                std::string event = "event: error\ndata: " + error_data.dump() + "\n\n";
                sink.write(event.c_str(), event.size());
            } catch (const std::exception& e) {
                std::string error_msg = e.what();
                if (error_msg != "Download cancelled") {
                    nlohmann::json error_data = {{"error", error_msg}};
                    std::string event = "event: error\ndata: " + error_data.dump() + "\n\n";
                    sink.write(event.c_str(), event.size());
                }
            }

            sink.done();
            return false;
        });
}



void Server::handle_downloads(const httplib::Request&, httplib::Response& res) {
    nlohmann::json response = nlohmann::json::array();
    std::vector<std::shared_ptr<DownloadJob>> expired_jobs;
    const auto now = std::chrono::steady_clock::now();

    {
        std::lock_guard<std::mutex> lock(downloads_mutex_);
        for (auto it = download_jobs_.begin(); it != download_jobs_.end();) {
            const auto& job = it->second;
            if (is_download_job_visible(job)) {
                response.push_back(download_job_to_json(job));
                ++it;
                continue;
            }

            const bool expired_terminal = job &&
                !job->running &&
                job->status == "completed" &&
                job->terminal_since.time_since_epoch().count() > 0 &&
                now - job->terminal_since >= DOWNLOAD_TERMINAL_VISIBILITY;

            if (expired_terminal) {
                expired_jobs.push_back(job);
                it = download_jobs_.erase(it);
            } else {
                ++it;
            }
        }
    }

    for (auto& job : expired_jobs) {
        join_download_job(job);
    }

    res.set_content(response.dump(), "application/json");
}


void Server::handle_download_control(const httplib::Request& req, httplib::Response& res) {
    try {
        auto request_json = nlohmann::json::parse(req.body);
        std::string id = request_json.value("id", "");
        std::string action = request_json.value("action", "");

        if (id.empty() || action.empty()) {
            res.status = 400;
            res.set_content("{\"error\": \"Both 'id' and 'action' are required\"}", "application/json");
            return;
        }

        nlohmann::json response_json;
        std::shared_ptr<DownloadJob> job_to_join;

        {
            std::lock_guard<std::mutex> lock(downloads_mutex_);
            auto it = download_jobs_.find(id);
            if (it == download_jobs_.end()) {
                if (action == "remove") {
                    res.set_content("{\"status\": \"ok\", \"missing\": true}", "application/json");
                    return;
                }
                res.status = 404;
                res.set_content("{\"error\": \"Download not found\"}", "application/json");
                return;
            }

            auto job = it->second;
            if (action == "pause" || action == "cancel") {
                const bool terminal = job->status == "completed" ||
                    job->status == "cancelled" ||
                    job->status == "error";

                if (!terminal) {
                    job->cancel_requested = true;
                    job->cancel_action = action;
                    job->status = action == "cancel" ? "cancelled" : "paused";
                    // Paused jobs remain visible until resumed/removed. A cancel
                    // request for an already-stopped job has no worker that will
                    // later stamp terminal_since, so start the terminal visibility
                    // window here to avoid leaving a stale hidden registry entry.
                    job->terminal_since = (action == "cancel" && !job->running)
                        ? std::chrono::steady_clock::now()
                        : std::chrono::steady_clock::time_point{};
                }

                response_json = download_job_to_json(job);
            } else if (action == "remove") {
                if (job->running) {
                    // A remove request must not make the job disappear while the
                    // worker may still hold file handles. Convert it to a cancel
                    // request and keep the job visible until running=false.
                    job->cancel_requested = true;
                    job->cancel_action = "cancel";
                    if (job->status != "completed" && job->status != "error") {
                        job->status = "cancelled";
                        job->terminal_since = std::chrono::steady_clock::time_point{};
                    }
                    response_json = download_job_to_json(job);
                } else {
                    job_to_join = job;
                    download_jobs_.erase(it);
                    response_json = {{"status", "ok"}};
                }
            } else {
                res.status = 400;
                res.set_content("{\"error\": \"Unsupported download action\"}", "application/json");
                return;
            }
        }

        join_download_job(job_to_join);
        res.set_content(response_json.dump(), "application/json");
    } catch (const std::exception& e) {
        res.status = 400;
        nlohmann::json error = {{"error", e.what()}};
        res.set_content(error.dump(), "application/json");
    }
}

void Server::cancel_download_jobs() {
    std::vector<std::shared_ptr<DownloadJob>> jobs;
    {
        std::lock_guard<std::mutex> lock(downloads_mutex_);
        for (auto& [id, job] : download_jobs_) {
            job->cancel_requested = true;
            job->cancel_action = "cancel";
            jobs.push_back(job);
        }
    }

    for (auto& job : jobs) {
        join_download_job(job);
    }
}


// ============================================================================
// Backend management endpoints
// ============================================================================

void Server::handle_install(const httplib::Request& req, httplib::Response& res) {
    try {
        auto request_json = nlohmann::json::parse(req.body);

        // Cloud install branch. Cloud providers don't have a binary to fetch —
        // "installing" one means registering its OpenAI-compat base URL with
        // the in-process CloudProviderRegistry so that ModelManager can
        // discover its catalog as soon as the matching API key is supplied
        // (env var or POST /v1/cloud/auth). Shape:
        //   {backend: "cloud", provider: "fireworks",
        //    base_url: "https://api.fireworks.ai/inference/v1",
        //    allow_insecure_http: false,
        //    auth_header_name / auth_header_prefix / wire_format: see
        //      CloudProviderRegistry::Record,
        //    api_key: "..."}
        //
        // Every field after base_url is optional and applied only when
        // present, so a re-install that omits one (the desktop add-provider
        // form posts just backend / provider / base_url / api_key) keeps the
        // stored value.
        if (request_json.value("backend", "") == "cloud") {
            const std::string provider = request_json.value("provider", "");
            const std::string base_url = request_json.value("base_url", "");
            const std::string api_key = request_json.value("api_key", "");

            auto reject = [&](const std::string& message) {
                res.status = 400;
                nlohmann::json error = {{"error", {
                    {"message", message},
                    {"type", "invalid_request_error"}}}};
                res.set_content(error.dump(), "application/json");
            };
            if (provider.empty() || base_url.empty()) {
                reject("Cloud install requires 'provider' and 'base_url' string fields");
                return;
            }
            if (auto err = CloudProviderRegistry::validate_provider_name(provider); !err.empty()) {
                reject(err);
                return;
            }
            if (auto err = CloudProviderRegistry::validate_base_url(base_url); !err.empty()) {
                reject(err);
                return;
            }

            CloudProviderRegistry::InstallOptions install_options;
            auto read_validated_field = [&](const char* field,
                                            std::string (*validate)(const std::string&),
                                            std::optional<std::string>& out) {
                if (!request_json.contains(field)) return true;
                if (!request_json[field].is_string()) {
                    reject(std::string(field) + " must be a string when provided");
                    return false;
                }
                auto value = request_json[field].get<std::string>();
                if (auto err = validate(value); !err.empty()) {
                    reject(err);
                    return false;
                }
                out = std::move(value);
                return true;
            };
            if (!read_validated_field("auth_header_name",
                                      CloudProviderRegistry::validate_auth_header_name,
                                      install_options.auth_header_name) ||
                !read_validated_field("auth_header_prefix",
                                      CloudProviderRegistry::validate_auth_header_prefix,
                                      install_options.auth_header_prefix) ||
                !read_validated_field("wire_format",
                                      CloudProviderRegistry::validate_wire_format,
                                      install_options.wire_format)) {
                return;
            }
            if (request_json.contains("allow_insecure_http")) {
                if (!request_json["allow_insecure_http"].is_boolean()) {
                    reject("allow_insecure_http must be a boolean when provided");
                    return;
                }
                install_options.allow_insecure_http =
                    request_json["allow_insecure_http"].get<bool>();
            }
            // The http:// opt-in gate below reflects the effective state: an
            // omitted flag on a re-install keeps whatever was stored.
            const bool allow_insecure_http =
                install_options.allow_insecure_http.value_or(
                    cloud_registry_->allow_insecure_http_for(provider));

            const auto env_state = cloud_registry_->auth_state(provider);
            if (CloudProviderRegistry::is_http_base_url(base_url) &&
                !allow_insecure_http &&
                (!api_key.empty() || env_state.env_var_set)) {
                res.status = 400;
                nlohmann::json error = {{"error", {
                    {"message", "Cloud provider '" + provider + "' uses http://. "
                                "Set allow_insecure_http=true to explicitly opt in before "
                                "storing or using an API key over plaintext HTTP."},
                    {"type", "invalid_request_error"},
                    {"code", "insecure_http_requires_opt_in"}}}};
                res.set_content(error.dump(), "application/json");
                return;
            }
            LOG(INFO, "Server") << "Installing cloud provider '" << provider
                                  << "' with base_url " << base_url << std::endl;
            cloud_registry_->install(provider, base_url, install_options);
            persist_cloud_providers();

            // Best-effort optional auth: if api_key was supplied, treat this
            // as "install + auth in one shot". Honors the env-wins rule: if
            // LEMONADE_<P>_API_KEY is set, we don't store the runtime key and
            // we don't error — install still succeeds, env var still wins.
            bool runtime_key_stored = false;
            if (!api_key.empty()) {
                runtime_key_stored = cloud_registry_->set_runtime_key(provider, api_key);
            }

            // Best-effort discovery. Empty result is fine — install means
            // "registered", not "verified-and-non-empty". The client can
            // call /v1/system-info later to see how many models showed up.
            size_t models_after = model_manager_->refresh_cloud_models(provider);
            const auto state = cloud_registry_->auth_state(provider);
            const auto auth_header = cloud_registry_->auth_header_for(provider);

            nlohmann::json response = {
                {"status", "success"},
                {"backend", "cloud"},
                {"provider", provider},
                {"base_url", cloud_registry_->base_url_for(provider)},
                {"allow_insecure_http", cloud_registry_->allow_insecure_http_for(provider)},
                {"auth_header_name", auth_header.name},
                {"auth_header_prefix", auth_header.prefix},
                {"wire_format", cloud_registry_->wire_format_for(provider)},
                {"models_discovered", models_after},
                {"auth_state", {
                    {"env_var_set", state.env_var_set},
                    {"runtime_key_set", state.runtime_key_set}
                }}
            };
            std::vector<std::string> warnings = CloudProviderRegistry::base_url_warnings(
                cloud_registry_->base_url_for(provider),
                state.env_var_set || state.runtime_key_set);
            if (!api_key.empty() && !runtime_key_stored) {
                warnings.push_back(CloudProviderRegistry::env_var_name(provider) +
                    " is set; supplied api_key was ignored.");
            }
            attach_warnings(response, warnings);
            res.set_content(response.dump(), "application/json");
            return;
        }

        std::string recipe = request_json.value("recipe", "");
        std::string backend = request_json.value("backend", "");
        bool stream = request_json.value("stream", false);
        bool subscribe = request_json.value("subscribe", true);
        bool force = request_json.value("force", false);

        if (recipe.empty() || backend.empty()) {
            res.status = 400;
            nlohmann::json error = {{"error", "Both 'recipe' and 'backend' are required"}};
            res.set_content(error.dump(), "application/json");
            return;
        }

        LOG(INFO, "Server") << "Installing backend: " << recipe << ":" << backend << std::endl;

        // Get fresh state before any checks
        SystemInfoCache::invalidate_recipes();

        // Check if this backend requires manual setup (e.g. FLM on Linux).
        // If so, return the action URL instead of attempting installation.
        json system_info = SystemInfoCache::get_system_info_with_cache();
        if (system_info.contains("recipes") &&
            system_info["recipes"].contains(recipe) &&
            system_info["recipes"][recipe].contains("backends") &&
            system_info["recipes"][recipe]["backends"].contains(backend)) {
            const auto& backend_info = system_info["recipes"][recipe]["backends"][backend];
            std::string state = backend_info.value("state", "unsupported");
            std::string message = backend_info.value("message", "Backend is not supported on this system.");
            std::string action = backend_info.value("action", "");

            if (state == "unsupported" && !force) {
                res.status = 400;
                nlohmann::json error = {
                    {"error", "Cannot install " + recipe + ":" + backend + " on this system: " + message},
                    {"recipe", recipe},
                    {"backend", backend}
                };
                res.set_content(error.dump(), "application/json");
                return;
            }

            if (action.find(".html") != std::string::npos) {
                auto url_pos = action.find("https://");
                if (url_pos != std::string::npos) {
                    nlohmann::json response = {
                        {"action", action.substr(url_pos)},
                        {"recipe", recipe},
                        {"backend", backend}
                    };
                    res.set_content(response.dump(), "application/json");
                    return;
                }
            }
        }

        if (stream) {
            auto operation = [this, recipe, backend, force](DownloadProgressCallback progress_cb) {
                backend_manager_->install_backend(recipe, backend, force, progress_cb);
                SystemInfoCache::invalidate_recipes();
                model_manager_->invalidate_models_cache();
            };

            if (!subscribe) {
                // Server-owned mode for desktop UI reload/new-tab recovery.
                // Legacy streamed /install behavior is unchanged.
                const std::string display_name = recipe + ":" + backend;
                auto job = start_download_job("backend:" + display_name, "backend", display_name, operation);
                nlohmann::json response;
                {
                    std::lock_guard<std::mutex> lock(downloads_mutex_);
                    response = download_job_to_json(job);
                }
                res.set_content(response.dump(), "application/json");
                return;
            }

            stream_download_operation(res, operation);
        } else {
            backend_manager_->install_backend(recipe, backend, force);
            SystemInfoCache::invalidate_recipes();
            model_manager_->invalidate_models_cache();
            nlohmann::json response = {
                {"status", "success"},
                {"recipe", recipe},
                {"backend", backend}
            };
            res.set_content(response.dump(), "application/json");
        }

    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "ERROR in handle_install: " << e.what() << std::endl;
        res.status = 500;
        nlohmann::json error = {{"error", e.what()}};
        res.set_content(error.dump(), "application/json");
    }
}

void Server::handle_install_dry_run(const httplib::Request& req, httplib::Response& res) {
    // Resolve the backend asset URL that install_backend() would download for a
    // given recipe/backend, optionally for a mocked GPU arch, without fetching
    // any bytes — so a 404 from a stale target-name or version pin can be caught
    // for any arch without that GPU present.
    std::string requested_arch;
    try {
        auto request_json = nlohmann::json::parse(req.body);

        const std::string recipe = request_json.value("recipe", "");
        const std::string backend = request_json.value("backend", "");
        requested_arch = request_json.value("arch", "");

        if (recipe.empty() || backend.empty()) {
            res.status = 400;
            nlohmann::json error = {{"error", "Both 'recipe' and 'backend' are required"}};
            res.set_content(error.dump(), "application/json");
            return;
        }

        if (!requested_arch.empty()) {
            SystemInfo::set_rocm_arch_override(requested_arch);
        }

        auto params = backend_manager_->get_install_params(recipe, backend);

        // Clear the override as soon as resolution is done so it cannot leak to
        // any later work on this thread.
        SystemInfo::set_rocm_arch_override("");

        const std::string url = "https://github.com/" + params.repo +
                                "/releases/download/" + params.version + "/" +
                                params.filename;

        bool supports_split_archive = false;
        if (auto* spec = backends::try_get_spec_for_recipe(recipe)) {
            supports_split_archive = spec->supports_split_archive;
        }

        bool supported = requested_arch.empty()
            ? true
            : SystemInfo::backend_supports_arch(recipe, backend, requested_arch);

        nlohmann::json response = {
            {"recipe", recipe},
            {"backend", backend},
            {"arch", requested_arch},
            {"repo", params.repo},
            {"version", params.version},
            {"filename", params.filename},
            {"url", url},
            {"supports_split_archive", supports_split_archive},
            {"supported", supported}
        };
        res.set_content(response.dump(), "application/json");
    } catch (const std::exception& e) {
        SystemInfo::set_rocm_arch_override("");
        res.status = 500;
        nlohmann::json error = {{"error", e.what()}};
        if (!requested_arch.empty()) {
            error["arch"] = requested_arch;
        }
        res.set_content(error.dump(), "application/json");
    }
}

void Server::handle_uninstall(const httplib::Request& req, httplib::Response& res) {
    try {
        auto request_json = nlohmann::json::parse(req.body);

        // Cloud uninstall: drop the provider record + its runtime key +
        // every discovered model. Idempotent — uninstalling an unknown
        // provider returns 404 (matching the symmetric install error shape)
        // rather than silently succeeding, so CI scripts can tell which
        // case happened.
        if (request_json.value("backend", "") == "cloud") {
            const std::string provider = request_json.value("provider", "");
            if (provider.empty()) {
                res.status = 400;
                nlohmann::json error = {{"error", {
                    {"message", "Cloud uninstall requires 'provider' string field"},
                    {"type", "invalid_request_error"}}}};
                res.set_content(error.dump(), "application/json");
                return;
            }
            // Unload any loaded cloud-recipe models for this provider before
            // erasing them from the cache. Router::unload_model expects a
            // model_name, not a recipe filter — walk the loaded list manually.
            auto loaded = router_->get_all_loaded_models();
            for (const auto& m : loaded) {
                if (m.value("recipe", "") == "cloud" &&
                    m.value("cloud_provider", "") == provider) {
                    router_->unload_model(m.value("model_name", ""));
                }
            }
            bool removed = cloud_registry_->uninstall(provider);
            size_t evicted = model_manager_->evict_cloud_models(provider);
            if (!removed) {
                res.status = 404;
                nlohmann::json error = {{"error", {
                    {"message", "Cloud provider '" + provider + "' is not installed"},
                    {"type", "invalid_request_error"}}}};
                res.set_content(error.dump(), "application/json");
                return;
            }
            persist_cloud_providers();
            nlohmann::json response = {
                {"status", "success"},
                {"backend", "cloud"},
                {"provider", provider},
                {"models_evicted", evicted}
            };
            res.set_content(response.dump(), "application/json");
            return;
        }

        std::string recipe = request_json.value("recipe", "");
        std::string backend = request_json.value("backend", "");

        if (recipe.empty() || backend.empty()) {
            res.status = 400;
            nlohmann::json error = {{"error", "Both 'recipe' and 'backend' are required"}};
            res.set_content(error.dump(), "application/json");
            return;
        }

        LOG(INFO, "Server") << "Uninstalling backend: " << recipe << ":" << backend << std::endl;

        // Check if any loaded models use this recipe+backend and unload them first
        auto loaded_models = router_->get_all_loaded_models();
        std::string backend_option_key = recipe + "_backend";
        for (const auto& model : loaded_models) {
            if (model.value("recipe", "") == recipe) {
                // Check if the model's backend matches the one being uninstalled
                std::string model_backend;
                if (model.contains("recipe_options") && model["recipe_options"].contains(backend_option_key)) {
                    model_backend = model["recipe_options"].value(backend_option_key, "");
                }
                if (!model_backend.empty() && model_backend != backend) {
                    continue;  // Different backend, skip
                }
                std::string model_name = model.value("model_name", "");
                LOG(INFO, "Server") << "Unloading model " << model_name
                          << " before uninstalling " << recipe << ":" << backend << std::endl;
                router_->unload_model(model_name);
            }
        }

        backend_manager_->uninstall_backend(recipe, backend);

        SystemInfoCache::invalidate_recipes();
        model_manager_->invalidate_models_cache();

        nlohmann::json response = {
            {"status", "success"},
            {"recipe", recipe},
            {"backend", backend}
        };
        res.set_content(response.dump(), "application/json");

    } catch (const std::exception& e) {
        LOG(ERROR, "Server") << "ERROR in handle_uninstall: " << e.what() << std::endl;
        res.status = 500;
        nlohmann::json error = {{"error", e.what()}};
        res.set_content(error.dump(), "application/json");
    }
}

void Server::enrich_recipes(json& recipes) {
    if (!backend_manager_) return;

    for (auto& [recipe_name, recipe_info] : recipes.items()) {
        if (!recipe_info.contains("backends")) continue;
        for (auto& [backend_name, backend_info] : recipe_info["backends"].items()) {
            try {
                auto enrichment = backend_manager_->get_backend_enrichment(recipe_name, backend_name);
                if (!enrichment.release_url.empty()) {
                    backend_info["release_url"] = enrichment.release_url;
                }
                if (!enrichment.download_filename.empty()) {
                    backend_info["download_filename"] = enrichment.download_filename;
                }
                if (!backend_info.contains("version") || backend_info["version"].get<std::string>().empty()) {
                    if (!enrichment.version.empty()) {
                        backend_info["version"] = enrichment.version;
                    }
                }
            } catch (...) {}
        }
    }
}

} // namespace lemon
