llm-simple-router 0.8.0 → 0.9.0

package/config/recommended-providers.json

@@ -7,14 +7,20 @@
         "presetName": "deepseek",
         "apiType": "anthropic",
         "baseUrl": "https://api.deepseek.com/anthropic",
-        "models": ["deepseek-chat", "deepseek-reasoner"]
+        "models": [
+          "deepseek-v4-flash",
+          "deepseek-v4-pro"
+        ]
       },
       {
         "plan": "OpenAI",
         "presetName": "deepseek-openai",
         "apiType": "openai",
         "baseUrl": "https://api.deepseek.com",
-        "models": ["deepseek-chat", "deepseek-reasoner"]
+        "models": [
+          "deepseek-v4-flash",
+          "deepseek-v4-pro"
+        ]
       }
     ]
   },
@@ -86,7 +92,12 @@
         "presetName": "zhipu-coding-plan",
         "apiType": "anthropic",
         "baseUrl": "https://open.bigmodel.cn/api/anthropic",
-        "models": ["glm-5.1", "glm-5", "glm-4.7", "glm-4.5-air"]
+        "models": [
+          "glm-5.1",
+          "glm-5",
+          "glm-4.7",
+          "glm-4.5-air"
+        ]
       },
       {
         "plan": "API",
@@ -105,14 +116,17 @@
     ]
   },
   {
-    "group": "KIMI",
+    "group": "月之暗面",
     "presets": [
       {
         "plan": "Coding Plan",
         "presetName": "kimi-coding-plan",
         "apiType": "anthropic",
         "baseUrl": "https://api.kimi.com/coding",
-        "models": ["kimi-for-coding", "kimi-k2.5"]
+        "models": [
+          "kimi-for-coding",
+          "kimi-k2.5"
+        ]
       },
       {
         "plan": "API",
@@ -137,7 +151,9 @@
         "presetName": "minimax-token-plan",
         "apiType": "anthropic",
         "baseUrl": "https://api.minimaxi.com/anthropic",
-        "models": ["MiniMax-M2.7"]
+        "models": [
+          "MiniMax-M2.7"
+        ]
       },
       {
         "plan": "API",
@@ -258,7 +274,12 @@
         "presetName": "opencode-go-anthropic",
         "apiType": "anthropic",
         "baseUrl": "https://opencode.ai/zen/go/v1/messages",
-        "models": ["deepseek-v4-pro", "deepseek-v4-flash", "minimax-m2.7", "minimax-m2.5"]
+        "models": [
+          "deepseek-v4-pro",
+          "deepseek-v4-flash",
+          "minimax-m2.7",
+          "minimax-m2.5"
+        ]
       }
     ]
   },
@@ -270,7 +291,10 @@
         "presetName": "stepfun-step-plan",
         "apiType": "anthropic",
         "baseUrl": "https://api.stepfun.com/step_plan",
-        "models": ["step-3.5-flash-2603", "step-3.5-flash"]
+        "models": [
+          "step-3.5-flash-2603",
+          "step-3.5-flash"
+        ]
       },
       {
         "plan": "API",
@@ -288,4 +312,4 @@
       }
     ]
   }
-]
+]

package/config/recommended-retry-rules.json

@@ -1,10 +1,11 @@
 [
-  { "name": "429 Too Many Requests", "status_code": 429, "body_pattern": ".*", "retry_strategy": "exponential", "retry_delay_ms": 5000, "max_retries": 10, "max_delay_ms": 60000 },
-  { "name": "503 Service Unavailable", "status_code": 503, "body_pattern": ".*", "retry_strategy": "exponential", "retry_delay_ms": 5000, "max_retries": 10, "max_delay_ms": 60000 },
-  { "name": "ZAI 网络错误 (code 1234)", "status_code": 400, "body_pattern": "\"type\"\\s*:\\s*\"error\".*\"code\"\\s*:\\s*\"1234\"", "retry_strategy": "exponential", "retry_delay_ms": 5000, "max_retries": 10, "max_delay_ms": 60000 },
-  { "name": "ZAI 临时不可用", "status_code": 400, "body_pattern": "\"type\"\\s*:\\s*\"error\".*请稍后重试", "retry_strategy": "exponential", "retry_delay_ms": 5000, "max_retries": 10, "max_delay_ms": 60000 },
-  { "name": "ZAI 操作失败 (code 500)", "status_code": 400, "body_pattern": "\"type\"\\s*:\\s*\"error\".*\"code\"\\s*:\\s*\"500\"", "retry_strategy": "exponential", "retry_delay_ms": 5000, "max_retries": 10, "max_delay_ms": 60000 },
-  { "name": "ZAI 速率限制 (HTTP 200, code 1302)", "status_code": 200, "body_pattern": "\"error\".*\"code\"\\s*:\\s*\"1302\"", "retry_strategy": "exponential", "retry_delay_ms": 5000, "max_retries": 10, "max_delay_ms": 60000 },
-  { "name": "ZAI SSE 错误 (HTTP 200, code 500)", "status_code": 200, "body_pattern": "\"error\".*\"code\"\\s*:\\s*\"500\"", "retry_strategy": "exponential", "retry_delay_ms": 5000, "max_retries": 10, "max_delay_ms": 60000 },
-  { "name": "ZAI SSE 错误 (HTTP 200, code 1234)", "status_code": 200, "body_pattern": "\"error\".*\"code\"\\s*:\\s*\"1234\"", "retry_strategy": "exponential", "retry_delay_ms": 5000, "max_retries": 10, "max_delay_ms": 60000 }
+  { "name": "429 Too Many Requests", "status_code": 429, "body_pattern": ".*", "retry_strategy": "exponential", "retry_delay_ms": 5000, "max_retries": 10, "max_delay_ms": 60000, "providers": [] },
+  { "name": "503 Service Unavailable", "status_code": 503, "body_pattern": ".*", "retry_strategy": "exponential", "retry_delay_ms": 5000, "max_retries": 10, "max_delay_ms": 60000, "providers": [] },
+  { "name": "ZAI 网络错误 (code 1234)", "status_code": 400, "body_pattern": "\"type\"\\s*:\\s*\"error\".*\"code\"\\s*:\\s*\"1234\"", "retry_strategy": "exponential", "retry_delay_ms": 5000, "max_retries": 10, "max_delay_ms": 60000, "providers": ["智谱"] },
+  { "name": "ZAI 临时不可用", "status_code": 400, "body_pattern": "\"type\"\\s*:\\s*\"error\".*请稍后重试", "retry_strategy": "exponential", "retry_delay_ms": 5000, "max_retries": 10, "max_delay_ms": 60000, "providers": ["智谱"] },
+  { "name": "ZAI 操作失败 (code 500)", "status_code": 400, "body_pattern": "\"type\"\\s*:\\s*\"error\".*\"code\"\\s*:\\s*\"500\"", "retry_strategy": "exponential", "retry_delay_ms": 5000, "max_retries": 10, "max_delay_ms": 60000, "providers": ["智谱"] },
+  { "name": "ZAI 速率限制 (HTTP 200, code 1302)", "status_code": 200, "body_pattern": "\"error\".*\"code\"\\s*:\\s*\"1302\"", "retry_strategy": "exponential", "retry_delay_ms": 5000, "max_retries": 10, "max_delay_ms": 60000, "providers": ["智谱"] },
+  { "name": "ZAI SSE 错误 (HTTP 200, code 500)", "status_code": 200, "body_pattern": "\"error\".*\"code\"\\s*:\\s*\"500\"", "retry_strategy": "exponential", "retry_delay_ms": 5000, "max_retries": 10, "max_delay_ms": 60000, "providers": ["智谱"] },
+  { "name": "ZAI SSE 错误 (HTTP 200, code 1234)", "status_code": 200, "body_pattern": "\"error\".*\"code\"\\s*:\\s*\"1234\"", "retry_strategy": "exponential", "retry_delay_ms": 5000, "max_retries": 10, "max_delay_ms": 60000, "providers": ["智谱"] },
+  { "name": "KIMI 401 认证错误", "status_code": 401, "body_pattern": ".*authentication_error.*", "retry_strategy": "exponential", "retry_delay_ms": 5000, "max_retries": 3, "max_delay_ms": 60000, "providers": ["月之暗面"] }
 ]

package/dist/admin/providers.js

@@ -58,9 +58,11 @@ function cascadeProviderDisable(db, providerId) {
     return result;
 }
 function extractModelOverrides(models) {
-    const names = models.map(m => typeof m === "string" ? m : m.name);
+    const entries = models.map(m => typeof m === "string"
+        ? { name: m, patches: [] }
+        : { name: m.name, context_window: m.context_window, patches: m.patches ?? [] });
     const overrides = models.filter((m) => typeof m !== "string" && m.context_window != null);
-    return { names, overrides };
+    return { entries, overrides };
 }
 const API_KEY_PREVIEW_PREFIX_LEN = 4;
 const PROVIDER_NAME_RE = /^[a-zA-Z0-9_-]+$/;
@@ -71,7 +73,7 @@ const CreateProviderSchema = Type.Object({
     api_key: Type.String({ minLength: 1 }),
     models: Type.Optional(Type.Array(Type.Union([
         Type.String(),
-        Type.Object({ name: Type.String(), context_window: Type.Optional(Type.Number()) })
+        Type.Object({ name: Type.String(), context_window: Type.Optional(Type.Number()), patches: Type.Optional(Type.Array(Type.String())) })
     ]))),
     is_active: Type.Optional(Type.Number()),
     max_concurrency: Type.Optional(Type.Integer({ minimum: 0 })),
@@ -86,7 +88,7 @@ const UpdateProviderSchema = Type.Object({
     api_key: Type.Optional(Type.String({ minLength: 1 })),
     models: Type.Optional(Type.Array(Type.Union([
         Type.String(),
-        Type.Object({ name: Type.String(), context_window: Type.Optional(Type.Number()) })
+        Type.Object({ name: Type.String(), context_window: Type.Optional(Type.Number()), patches: Type.Optional(Type.Array(Type.String())) })
     ]))),
     is_active: Type.Optional(Type.Number()),
     max_concurrency: Type.Optional(Type.Integer({ minimum: 0 })),
@@ -100,7 +102,7 @@ export const adminProviderRoutes = (app, options, done) => {
         const encryptionKey = getSetting(db, "encryption_key");
         const providers = getAllProviders(db);
         return reply.send(providers.map((s) => {
-            const modelNames = parseModels(s.models || "[]");
+            const modelEntries = parseModels(s.models || "[]");
             const overrides = new Map(getModelInfoForProvider(db, s.id).map(m => [m.model_name, m.context_window]));
             return {
                 id: s.id,
@@ -108,7 +110,7 @@ export const adminProviderRoutes = (app, options, done) => {
                 api_type: s.api_type,
                 base_url: s.base_url,
                 api_key: s.api_key ? decrypt(s.api_key, encryptionKey) : "",
-                models: buildModelInfoList(modelNames, overrides),
+                models: buildModelInfoList(modelEntries, overrides),
                 is_active: s.is_active,
                 max_concurrency: s.max_concurrency,
                 queue_timeout_ms: s.queue_timeout_ms,
@@ -130,7 +132,7 @@ export const adminProviderRoutes = (app, options, done) => {
             return reply.code(HTTP_CONFLICT).send(apiError(API_CODE.CONFLICT_NAME, `Provider 名称 '${body.name}' 已存在`));
         }
         const encryptedKey = encrypt(body.api_key, getSetting(db, "encryption_key"));
-        const { names: normalizedModels, overrides: contextOverrides } = extractModelOverrides((body.models ?? []));
+        const { entries: normalizedModels, overrides: contextOverrides } = extractModelOverrides((body.models ?? []));
         const isAdaptiveEnabled = body.adaptive_enabled ?? 0;
         const id = createProvider(db, {
             name: body.name,
@@ -190,8 +192,8 @@ export const adminProviderRoutes = (app, options, done) => {
         if (body.is_active !== undefined)
             fields.is_active = body.is_active;
         if (body.models !== undefined) {
-            const { names, overrides } = extractModelOverrides(body.models);
-            fields.models = JSON.stringify(names);
+            const { entries, overrides } = extractModelOverrides(body.models);
+            fields.models = JSON.stringify(entries);
             if (overrides.length > 0) {
                 setModelInfoForProvider(db, id, overrides.map(o => ({ model_name: o.name, context_window: o.context_window })));
             }

package/dist/admin/proxy-enhancement.js

@@ -4,6 +4,7 @@ const UpdateProxyEnhancementSchema = Type.Object({
     claude_code_enabled: Type.Boolean(),
     tool_call_loop_enabled: Type.Boolean(),
     stream_loop_enabled: Type.Boolean(),
+    tool_round_limit_enabled: Type.Boolean(),
 });
 const SessionParamsSchema = Type.Object({
     keyId: Type.String(),
@@ -13,7 +14,7 @@ import { getSessionStates, getSessionHistory, } from "../db/session-states.js";
 export const adminProxyEnhancementRoutes = (app, options, done) => {
     const { db, stateRegistry } = options;
     app.get("/admin/api/proxy-enhancement", async (_request, reply) => {
-        const config = stateRegistry?.getEnhancementConfig() ?? { claude_code_enabled: false, tool_call_loop_enabled: false, stream_loop_enabled: false };
+        const config = stateRegistry?.getEnhancementConfig() ?? { claude_code_enabled: false, tool_call_loop_enabled: false, stream_loop_enabled: false, tool_round_limit_enabled: true };
         return reply.send(config);
     });
     app.put("/admin/api/proxy-enhancement", { schema: { body: UpdateProxyEnhancementSchema } }, async (request, reply) => {
@@ -22,6 +23,7 @@ export const adminProxyEnhancementRoutes = (app, options, done) => {
             claude_code_enabled: body.claude_code_enabled,
             tool_call_loop_enabled: body.tool_call_loop_enabled,
             stream_loop_enabled: body.stream_loop_enabled,
+            tool_round_limit_enabled: body.tool_round_limit_enabled,
         };
         setSetting(db, "proxy_enhancement", JSON.stringify(config));
         return reply.send({ success: true });

package/dist/admin/quick-setup.d.ts

@@ -0,0 +1,13 @@
+import { FastifyPluginCallback } from "fastify";
+import Database from "better-sqlite3";
+import type { StateRegistry } from "../core/registry.js";
+import type { RequestTracker } from "../monitor/request-tracker.js";
+import type { AdaptiveConcurrencyController } from "../proxy/adaptive-controller.js";
+interface QuickSetupRoutesOptions {
+    db: Database.Database;
+    stateRegistry?: StateRegistry;
+    tracker?: RequestTracker;
+    adaptiveController?: AdaptiveConcurrencyController;
+}
+export declare const adminQuickSetupRoutes: FastifyPluginCallback<QuickSetupRoutesOptions>;
+export {};

package/dist/admin/quick-setup.js

@@ -0,0 +1,169 @@
+import { Type } from "@sinclair/typebox";
+import { createProvider } from "../db/providers.js";
+import { createMappingGroup, updateMappingGroup } from "../db/mappings.js";
+import { createRetryRule } from "../db/retry-rules.js";
+import { upsertTransformRule } from "../db/transform-rules.js";
+import { encrypt } from "../utils/crypto.js";
+import { getSetting } from "../db/settings.js";
+import { HTTP_CREATED, HTTP_BAD_REQUEST, HTTP_CONFLICT } from "./constants.js";
+import { API_CODE, apiError } from "./api-response.js";
+import { PROVIDER_CONCURRENCY_DEFAULTS } from "../db/providers.js";
+const PROVIDER_NAME_RE = /^[a-zA-Z0-9_-]+$/;
+const API_KEY_PREVIEW_MIN_LENGTH = 8;
+const API_KEY_PREVIEW_PREFIX_LEN = 4;
+const QuickSetupProviderSchema = Type.Object({
+    name: Type.String({ minLength: 1 }),
+    api_type: Type.Union([Type.Literal("openai"), Type.Literal("anthropic")]),
+    base_url: Type.String({ minLength: 1 }),
+    api_key: Type.String({ minLength: 1 }),
+    models: Type.Array(Type.Object({
+        name: Type.String(),
+        context_window: Type.Optional(Type.Number()),
+        patches: Type.Optional(Type.Array(Type.String())),
+    })),
+    concurrency_mode: Type.Optional(Type.Union([Type.Literal("auto"), Type.Literal("manual"), Type.Literal("none")])),
+    max_concurrency: Type.Optional(Type.Number()),
+    queue_timeout_ms: Type.Optional(Type.Number()),
+    max_queue_size: Type.Optional(Type.Number()),
+});
+const QuickSetupMappingSchema = Type.Object({
+    client_model: Type.String({ minLength: 1 }),
+    backend_model: Type.String({ minLength: 1 }),
+});
+const QuickSetupRetryRuleSchema = Type.Object({
+    name: Type.String({ minLength: 1 }),
+    status_code: Type.Number({ minimum: 100, maximum: 599 }),
+    body_pattern: Type.String({ minLength: 1 }),
+    retry_strategy: Type.Union([Type.Literal("fixed"), Type.Literal("exponential")]),
+    retry_delay_ms: Type.Number({ minimum: 100 }),
+    max_retries: Type.Number({ minimum: 0, maximum: 100 }),
+    max_delay_ms: Type.Number({ minimum: 100 }),
+});
+const QuickSetupTransformSchema = Type.Object({
+    inject_headers: Type.Optional(Type.Record(Type.String(), Type.String())),
+    request_defaults: Type.Optional(Type.Record(Type.String(), Type.Unknown())),
+    drop_fields: Type.Optional(Type.Array(Type.String())),
+});
+const QuickSetupSchema = Type.Object({
+    provider: QuickSetupProviderSchema,
+    mappings: Type.Array(QuickSetupMappingSchema),
+    retry_rules: Type.Array(QuickSetupRetryRuleSchema),
+    transform_rules: Type.Optional(QuickSetupTransformSchema),
+});
+export const adminQuickSetupRoutes = (app, options, done) => {
+    const { db, stateRegistry, tracker, adaptiveController } = options;
+    app.post("/admin/api/quick-setup", { schema: { body: QuickSetupSchema } }, async (request, reply) => {
+        const body = request.body;
+        // 1. Validate provider name
+        if (!PROVIDER_NAME_RE.test(body.provider.name)) {
+            return reply.code(HTTP_BAD_REQUEST).send(apiError(API_CODE.VALIDATION_FAILED, "Provider 名称仅允许英文大小写字母、数字、横线和下划线"));
+        }
+        // 2. Check no duplicate provider name
+        const existing = db.prepare("SELECT id FROM providers WHERE name = ?").get(body.provider.name);
+        if (existing) {
+            return reply.code(HTTP_CONFLICT).send(apiError(API_CODE.CONFLICT_NAME, `Provider 名称 '${body.provider.name}' 已存在`));
+        }
+        // 3. Validate retry rule body_pattern regex
+        for (const rule of body.retry_rules) {
+            try {
+                new RegExp(rule.body_pattern);
+            }
+            catch {
+                return reply.code(HTTP_BAD_REQUEST).send(apiError(API_CODE.INVALID_REGEX, `重试规则「${rule.name}」的 body_pattern 不是有效的正则表达式`));
+            }
+        }
+        // 4. Start transaction
+        const encryptionKey = getSetting(db, "encryption_key");
+        const createAll = db.transaction(() => {
+            // 5. Create provider with models JSON
+            const encryptedKey = encrypt(body.provider.api_key, encryptionKey);
+            const modelEntries = body.provider.models.map(m => ({
+                name: m.name,
+                ...(m.context_window != null ? { context_window: m.context_window } : {}),
+                ...(m.patches && m.patches.length > 0 ? { patches: m.patches } : {}),
+            }));
+            const adaptiveEnabled = body.provider.concurrency_mode === 'auto' ? 1 : 0;
+            const maxConcurrency = body.provider.max_concurrency ?? PROVIDER_CONCURRENCY_DEFAULTS.max_concurrency;
+            const queueTimeoutMs = body.provider.queue_timeout_ms ?? PROVIDER_CONCURRENCY_DEFAULTS.queue_timeout_ms;
+            const maxQueueSize = body.provider.max_queue_size ?? PROVIDER_CONCURRENCY_DEFAULTS.max_queue_size;
+            const providerId = createProvider(db, {
+                name: body.provider.name,
+                api_type: body.provider.api_type,
+                base_url: body.provider.base_url,
+                api_key: encryptedKey,
+                api_key_preview: body.provider.api_key.length > API_KEY_PREVIEW_MIN_LENGTH
+                    ? `${body.provider.api_key.slice(0, API_KEY_PREVIEW_PREFIX_LEN)}...${body.provider.api_key.slice(-API_KEY_PREVIEW_PREFIX_LEN)}`
+                    : "****",
+                models: JSON.stringify(modelEntries),
+                is_active: 1,
+                max_concurrency: maxConcurrency,
+                queue_timeout_ms: queueTimeoutMs,
+                max_queue_size: maxQueueSize,
+                adaptive_enabled: adaptiveEnabled,
+            });
+            // 6. Upsert mapping groups
+            for (const m of body.mappings) {
+                const existing = db.prepare('SELECT id FROM mapping_groups WHERE client_model = ?').get(m.client_model);
+                const ruleJson = JSON.stringify({
+                    targets: [{ backend_model: m.backend_model, provider_id: providerId }],
+                });
+                if (existing) {
+                    updateMappingGroup(db, existing.id, {
+                        client_model: m.client_model,
+                        rule: ruleJson,
+                    });
+                }
+                else {
+                    createMappingGroup(db, {
+                        client_model: m.client_model,
+                        rule: ruleJson,
+                    });
+                }
+            }
+            // 7. Create retry rules
+            for (const r of body.retry_rules) {
+                createRetryRule(db, {
+                    name: r.name,
+                    status_code: r.status_code,
+                    body_pattern: r.body_pattern,
+                    is_active: 1,
+                    retry_strategy: r.retry_strategy,
+                    retry_delay_ms: r.retry_delay_ms,
+                    max_retries: r.max_retries,
+                    max_delay_ms: r.max_delay_ms,
+                });
+            }
+            // 8. Create transform rules
+            if (body.transform_rules) {
+                upsertTransformRule(db, providerId, {
+                    inject_headers: body.transform_rules.inject_headers ?? null,
+                    request_defaults: body.transform_rules.request_defaults ?? null,
+                    drop_fields: body.transform_rules.drop_fields ?? null,
+                    is_active: 1,
+                });
+            }
+            return providerId;
+        });
+        // 8. Execute transaction
+        const providerId = createAll();
+        // 9. Sync concurrency state
+        const finalAdaptiveEnabled = body.provider.concurrency_mode === 'auto' ? 1 : 0;
+        const finalMaxConcurrency = body.provider.max_concurrency ?? PROVIDER_CONCURRENCY_DEFAULTS.max_concurrency;
+        const finalQueueTimeoutMs = body.provider.queue_timeout_ms ?? PROVIDER_CONCURRENCY_DEFAULTS.queue_timeout_ms;
+        const finalMaxQueueSize = body.provider.max_queue_size ?? PROVIDER_CONCURRENCY_DEFAULTS.max_queue_size;
+        adaptiveController?.syncProvider(providerId, {
+            adaptive_enabled: finalAdaptiveEnabled,
+            max_concurrency: finalMaxConcurrency,
+            queue_timeout_ms: finalQueueTimeoutMs,
+            max_queue_size: finalMaxQueueSize,
+        });
+        tracker?.updateProviderConfig(providerId, {
+            name: body.provider.name,
+            maxConcurrency: finalMaxConcurrency,
+            queueTimeoutMs: finalQueueTimeoutMs,
+            maxQueueSize: finalMaxQueueSize,
+        });
+        return reply.code(HTTP_CREATED).send({ success: true, provider_id: providerId });
+    });
+    done();
+};

package/dist/admin/recommended.js

@@ -7,7 +7,11 @@ export const adminRecommendedRoutes = (app, options, done) => {
     app.get("/admin/api/recommended/retry-rules", async (_req, reply) => {
         const rules = getRecommendedRetryRules();
         const existing = new Set(db.prepare("SELECT name FROM retry_rules").all().map((r) => r.name));
-        return reply.send(rules.filter((r) => !existing.has(r.name)));
+        // Return all rules with `exists` flag, so the frontend can show all and mark existing ones
+        return reply.send(rules.map(r => ({
+            ...r,
+            exists: existing.has(r.name),
+        })));
     });
     app.post("/admin/api/recommended/reload", async (_req, reply) => {
         reloadConfig();

package/dist/admin/routes.js

@@ -14,6 +14,7 @@ import { adminSettingsRoutes } from "./settings.js";
 import { adminRecommendedRoutes } from "./recommended.js";
 import { adminUsageRoutes } from "./usage.js";
 import { adminUpgradeRoutes } from "./upgrade.js";
+import { adminQuickSetupRoutes } from "./quick-setup.js";
 import { adminImportExportRoutes } from "./settings-import-export.js";
 import { adminTransformRuleRoutes } from "./transform-rules.js";
 import { adminScheduleRoutes } from "./schedules.js";
@@ -37,6 +38,7 @@ export const adminRoutes = (app, options, done) => {
     app.register(adminImportExportRoutes, { db: options.db, stateRegistry: options.stateRegistry, pluginRegistry: options.pluginRegistry });
     app.register(adminRecommendedRoutes, { db: options.db });
     app.register(adminUsageRoutes, { db: options.db });
+    app.register(adminQuickSetupRoutes, { db: options.db, stateRegistry: options.stateRegistry, tracker: options.tracker, adaptiveController: options.adaptiveController });
     app.register(adminUpgradeRoutes, { db: options.db, closeFn: options.closeFn ?? (async () => { }) });
     app.register(adminTransformRuleRoutes, { db: options.db, pluginRegistry: options.pluginRegistry });
     done();

package/dist/config/model-context.d.ts

@@ -1,10 +1,16 @@
 export interface ModelInfo {
     name: string;
     context_window: number | null;
+    patches: string[];
+}
+export interface ModelEntry {
+    name: string;
+    context_window?: number;
+    patches?: string[];
 }
 export declare const MODEL_CONTEXT_WINDOWS: Record<string, number>;
 export declare const DEFAULT_CONTEXT_WINDOW = 200000;
 export declare const OVERFLOW_THRESHOLD = 1000000;
 export declare function lookupContextWindow(modelName: string): number;
-export declare function parseModels(raw: string): string[];
-export declare function buildModelInfoList(modelNames: string[], overrides: Map<string, number>): ModelInfo[];
+export declare function parseModels(raw: string): ModelEntry[];
+export declare function buildModelInfoList(modelEntries: ModelEntry[], overrides: Map<string, number>): ModelInfo[];

package/dist/config/model-context.js

@@ -91,15 +91,27 @@ export function parseModels(raw) {
         const parsed = JSON.parse(raw);
         if (!Array.isArray(parsed))
             return [];
-        return parsed.map((item) => typeof item === 'string' ? item : item?.name ?? '').filter(Boolean);
+        return parsed.map((item) => {
+            if (typeof item === 'string') {
+                return item ? { name: item, patches: [] } : null;
+            }
+            const obj = item;
+            if (!obj || !obj.name)
+                return null;
+            return {
+                name: obj.name,
+                patches: obj.patches ?? [],
+            };
+        }).filter((e) => e !== null);
     }
     catch {
         return [];
     }
 }
-export function buildModelInfoList(modelNames, overrides) {
-    return modelNames.map(name => ({
-        name,
-        context_window: overrides.get(name) ?? lookupContextWindow(name),
+export function buildModelInfoList(modelEntries, overrides) {
+    return modelEntries.map(entry => ({
+        name: entry.name,
+        context_window: overrides.get(entry.name) ?? lookupContextWindow(entry.name),
+        patches: entry.patches ?? [],
     }));
 }

package/dist/config/recommended.d.ts

@@ -17,6 +17,7 @@ export interface RecommendedRetryRule {
     retry_delay_ms: number;
     max_retries: number;
     max_delay_ms: number;
+    providers?: string[];
 }
 export declare function loadRecommendedConfig(dir?: string): void;
 export declare function getRecommendedProviders(): ProviderGroup[];

package/dist/config/recommended.js

@@ -1,22 +1,18 @@
 import fs from 'fs';
 import path from 'path';
-let cachedProviders = [];
-let cachedRetryRules = [];
 let configDir = '';
 export function loadRecommendedConfig(dir) {
     configDir = dir ?? path.resolve(process.cwd(), 'config');
-    cachedProviders = loadJson('recommended-providers.json');
-    cachedRetryRules = loadJson('recommended-retry-rules.json');
 }
 export function getRecommendedProviders() {
-    return cachedProviders;
+    return loadJson('recommended-providers.json');
 }
 export function getRecommendedRetryRules() {
-    return cachedRetryRules;
-}
-export function reloadConfig() {
-    loadRecommendedConfig(configDir);
+    return loadJson('recommended-retry-rules.json');
 }
+// No-op: kept for backward compat (reload endpoint, upgrade flow)
+// Config is now always read from disk, no caching.
+export function reloadConfig() { }
 function loadJson(filename) {
     const filePath = path.join(configDir, filename);
     try {

package/dist/index.js

@@ -279,11 +279,17 @@ export async function buildApp(options) {
     const dbSizeMonitor = scheduleDbSizeMonitor(db, config.DB_PATH, {
         log: app.log,
     });
+    let closed = false;
     let close = async () => {
+        if (closed)
+            return;
+        closed = true;
         stopUpgradeChecker();
         logCleanup.stop();
         dbSizeMonitor.stop();
         tracker.stopPushInterval();
+        // 关闭所有 SSE 长连接，防止 app.close() 因 hijack 的连接无限等待
+        tracker.closeAllClients();
         modelState.clearAll();
         semaphoreManager.removeAll();
         const sessionTracker = container.resolve(SERVICE_KEYS.sessionTracker);
@@ -339,7 +345,22 @@ export async function main() {
         /* eslint-enable taste/no-silent-catch */
     });
     // 优雅关闭：SIGTERM（systemd/docker stop）和 SIGINT（Ctrl+C）
+    let isShuttingDown = false;
+    const GRACEFUL_SHUTDOWN_TIMEOUT_MS = 10_000;
     const shutdown = async (signal) => {
+        // 防止重复触发：多次 Ctrl+C 只执行一次关闭
+        if (isShuttingDown) {
+            app.log.info(`Received ${signal} again, already shutting down...`);
+            return;
+        }
+        isShuttingDown = true;
+        // 强制退出兜底：优雅关闭超过 N 秒则强制退出
+        const forceTimer = setTimeout(() => {
+            app.log.error("Graceful shutdown timed out, forcing exit");
+            process.exit(1);
+        }, GRACEFUL_SHUTDOWN_TIMEOUT_MS);
+        // 不阻止进程退出
+        forceTimer.unref();
         try {
             app.log.info(`Received ${signal}, shutting down gracefully...`);
             await close();
@@ -348,6 +369,7 @@ export async function main() {
         catch (err) {
             app.log.error({ err }, "Error during shutdown");
         }
+        clearTimeout(forceTimer);
         process.exit(0);
     };
     process.on("SIGTERM", () => shutdown("SIGTERM"));

package/dist/monitor/request-tracker.d.ts

@@ -52,6 +52,8 @@ export declare class RequestTracker {
     /** 向单个客户端发送当前活跃请求快照（保留 clientRequest 以便前端即时展示） */
     private sendInitialSnapshot;
     removeClient(res: ServerResponse): void;
+    /** 主动关闭所有 SSE 客户端连接，确保 app.close() 不会因长连接阻塞 */
+    closeAllClients(): void;
     startPushInterval(): void;
     stopPushInterval(): void;
     broadcast(event: string, data: unknown): void;

package/dist/monitor/request-tracker.js

@@ -188,6 +188,20 @@ export class RequestTracker {
     removeClient(res) {
         this.clients.delete(res);
     }
+    /** 主动关闭所有 SSE 客户端连接，确保 app.close() 不会因长连接阻塞 */
+    closeAllClients() {
+        const clients = [...this.clients];
+        this.clients.clear();
+        for (const client of clients) {
+            try {
+                if (!client.writableEnded)
+                    client.end();
+            }
+            catch {
+                // 忽略已关闭的连接
+            }
+        }
+    }
     // --- Push interval ---
     startPushInterval() {
         if (this.pushTimer)