openfox 2.0.24 → 2.0.26

package/dist/{chunk-XJ4SUDL7.js → chunk-QYP6MOB5.js}

@@ -1,132 +1,25 @@
+import {
+  getBackendCapabilities
+} from "./chunk-Z4FMBCJO.js";
+import {
+  ensureVersionPrefix
+} from "./chunk-HNCM3D7Y.js";
 import {
   buildNonStreamingCreateParams,
   buildStreamingCreateParams,
-  getBackendCapabilities,
-  getModelProfile,
   getThinking,
   mapFinishReason
-} from "./chunk-UUFEE7VR.js";
+} from "./chunk-ZJ4FP6RS.js";
+import {
+  LLMError,
+  OpenAIHttpClient
+} from "./chunk-IEDE6VK4.js";
 import {
   logger
 } from "./chunk-K44MW7JJ.js";
-
-// src/server/llm/url-utils.ts
-var VERSION_PREFIX_REGEX = /\/v\d+(\/|$)/;
-function hasVersionPrefix(url) {
-  return VERSION_PREFIX_REGEX.test(url);
-}
-function ensureVersionPrefix(url, defaultVersion = "/v1") {
-  if (hasVersionPrefix(url)) return url;
-  return `${url.replace(/\/+$/, "")}${defaultVersion}`;
-}
-function stripVersionPrefix(url) {
-  return url.replace(/\/v\d+\/?$/, "");
-}
-function buildModelsUrl(baseUrl) {
-  return `${ensureVersionPrefix(baseUrl)}/models`;
-}
-
-// src/server/utils/errors.ts
-var OpenFoxError = class extends Error {
-  constructor(message, code, details) {
-    super(message);
-    this.code = code;
-    this.details = details;
-    this.name = "OpenFoxError";
-  }
-};
-var SessionNotFoundError = class extends OpenFoxError {
-  constructor(sessionId) {
-    super(`Session not found: ${sessionId}`, "SESSION_NOT_FOUND", { sessionId });
-    this.name = "SessionNotFoundError";
-  }
-};
-var LLMError = class extends OpenFoxError {
-  constructor(message, details) {
-    super(message, "LLM_ERROR", details);
-    this.name = "LLMError";
-  }
-};
-
-// src/server/llm/http-client.ts
-import { Agent, setGlobalDispatcher } from "undici";
-var agent = new Agent({ allowH2: true });
-setGlobalDispatcher(agent);
-var OpenAIHttpClient = class {
-  baseURL;
-  apiKey;
-  constructor(options) {
-    this.baseURL = options.baseURL;
-    this.apiKey = options.apiKey;
-  }
-  async fetchChatCompletion(params, options) {
-    const url = `${this.baseURL}/chat/completions`;
-    const headers = {
-      "Content-Type": "application/json",
-      Authorization: `Bearer ${this.apiKey}`
-    };
-    const response = await fetch(url, {
-      method: "POST",
-      headers,
-      body: JSON.stringify(params),
-      signal: options?.signal ?? null
-    });
-    if (!response.ok) {
-      const errorText = await response.text();
-      throw new LLMError(`HTTP ${response.status}: ${errorText}`);
-    }
-    return response;
-  }
-  async createChatCompletion(params, options) {
-    const response = await this.fetchChatCompletion(params, options);
-    try {
-      const data = await response.json();
-      return data;
-    } catch (error) {
-      throw new LLMError(`Failed to parse response: ${error instanceof Error ? error.message : "Unknown error"}`);
-    }
-  }
-  createChatCompletionStream(params, options) {
-    const responsePromise = this.fetchChatCompletion(params, options);
-    async function* generate() {
-      const response = await responsePromise;
-      if (!response.body) {
-        throw new LLMError("No response body for streaming");
-      }
-      const reader = response.body.getReader();
-      const decoder = new TextDecoder();
-      let buffer = "";
-      try {
-        while (true) {
-          const { done, value } = await reader.read();
-          if (done) break;
-          buffer += decoder.decode(value, { stream: true });
-          const lines = buffer.split("\n");
-          buffer = lines.pop() || "";
-          for (const line of lines) {
-            const trimmed = line.trim();
-            if (!trimmed) continue;
-            if (trimmed.startsWith("data: ")) {
-              const data = trimmed.slice(6);
-              if (data === "[DONE]") {
-                return;
-              }
-              try {
-                const chunk = JSON.parse(data);
-                yield chunk;
-              } catch (error) {
-                logger.warn("Failed to parse SSE chunk", { data, error });
-              }
-            }
-          }
-        }
-      } finally {
-        reader.releaseLock();
-      }
-    }
-    return generate();
-  }
-};
+import {
+  getModelProfile
+} from "./chunk-V4IE7HJY.js";
 
 // src/server/llm/client.ts
 function createLLMClient(config, initialBackend = "unknown") {
@@ -141,7 +34,7 @@ function createLLMClient(config, initialBackend = "unknown") {
   let capabilities = getBackendCapabilities(backend);
   const reasoningEffort = config.llm.reasoningEffort;
   const thinkingField = config.llm.thinkingField;
-  const idleTimeout = config.llm.idleTimeout ?? 3e4;
+  const idleTimeout = config.llm.idleTimeout ?? 12e4;
   return {
     getModel() {
       return model;
@@ -176,7 +69,7 @@ function createLLMClient(config, initialBackend = "unknown") {
         reasoningEffort: request.reasoningEffort ?? reasoningEffort
       });
       try {
-        const resolvedEffort = request.reasoningEffort ?? reasoningEffort;
+        const resolvedEffort = request.skipClientReasoningEffort ? void 0 : request.reasoningEffort ?? reasoningEffort;
         const { params: createParams } = await buildNonStreamingCreateParams({
           model,
           request,
@@ -185,10 +78,14 @@ function createLLMClient(config, initialBackend = "unknown") {
           ...resolvedEffort ? { reasoningEffort: resolvedEffort } : {},
           ...thinkingField ? { thinkingField } : {}
         });
-        const response = await httpClient.createChatCompletion(createParams, {
-          signal: request.signal
-        });
-        const choice = response.choices[0];
+        const httpResponse = await httpClient.createChatCompletion(
+          createParams,
+          {
+            signal: request.signal
+          },
+          request.returnRaw
+        );
+        const choice = httpResponse.choices[0];
         if (!choice) {
           throw new LLMError("No completion choice returned");
         }
@@ -201,16 +98,17 @@ function createLLMClient(config, initialBackend = "unknown") {
           arguments: JSON.parse(tc.function.arguments)
         }));
         return {
-          id: response.id,
+          id: httpResponse.id,
           content,
           ...thinkingContent ? { thinkingContent } : {},
           ...toolCalls && toolCalls.length > 0 ? { toolCalls } : {},
           finishReason: mapFinishReason(choice.finish_reason),
           usage: {
-            promptTokens: response.usage?.prompt_tokens ?? 0,
-            completionTokens: response.usage?.completion_tokens ?? 0,
-            totalTokens: response.usage?.total_tokens ?? 0
-          }
+            promptTokens: httpResponse.usage?.prompt_tokens ?? 0,
+            completionTokens: httpResponse.usage?.completion_tokens ?? 0,
+            totalTokens: httpResponse.usage?.total_tokens ?? 0
+          },
+          ...httpResponse.raw ? { raw: httpResponse.raw } : {}
         };
       } catch (error) {
         logger.error("LLM complete error", { error: String(error) });
@@ -220,7 +118,7 @@ function createLLMClient(config, initialBackend = "unknown") {
       }
     },
     async *stream(request) {
-      const resolvedEffort = request.reasoningEffort ?? reasoningEffort;
+      const resolvedEffort = request.skipClientReasoningEffort ? void 0 : request.reasoningEffort ?? reasoningEffort;
       logger.debug("LLM stream request", {
         messageCount: request.messages.length,
         hasTools: !!request.tools?.length,
@@ -358,114 +256,7 @@ function createLLMClient(config, initialBackend = "unknown") {
   };
 }
 
-// src/server/llm/models.ts
-var modelCache = /* @__PURE__ */ new Map();
-var llmStatus = "unknown";
-var lastActiveUrl = null;
-var CACHE_TTL_MS = 3e4;
-function getCacheKey(url) {
-  return stripVersionPrefix(url);
-}
-async function detectModel(llmBaseUrl, retries = 3, silent = false) {
-  const cacheKey = getCacheKey(llmBaseUrl);
-  const now = Date.now();
-  const cached = modelCache.get(cacheKey);
-  if (cached && now - cached.timestamp < CACHE_TTL_MS) {
-    lastActiveUrl = cacheKey;
-    llmStatus = "connected";
-    return cached.model;
-  }
-  const url = buildModelsUrl(llmBaseUrl);
-  for (let attempt = 1; attempt <= retries; attempt++) {
-    try {
-      if (silent) {
-        logger.debug("Fetching models from LLM server", { url, attempt });
-      }
-      const response = await fetch(url, {
-        signal: AbortSignal.timeout(1e4)
-      });
-      if (!response.ok) {
-        if (silent) {
-          logger.debug("Failed to fetch models from LLM server", { status: response.status, attempt });
-        } else {
-          logger.warn("Failed to fetch models from LLM server", { status: response.status, attempt });
-        }
-        if (attempt < retries) {
-          await new Promise((r) => setTimeout(r, 1e3 * attempt));
-          continue;
-        }
-        llmStatus = "disconnected";
-        return cached?.model ?? null;
-      }
-      const data = await response.json();
-      if (data.data && data.data.length > 0) {
-        const modelData = data.data[0];
-        const modelId = modelData.id;
-        modelCache.set(cacheKey, {
-          model: modelId,
-          modelInfo: modelData,
-          timestamp: now
-        });
-        lastActiveUrl = cacheKey;
-        llmStatus = "connected";
-        if (silent) {
-          logger.debug("Detected LLM model", {
-            model: modelId,
-            maxLen: modelData.max_model_len,
-            root: modelData.root
-          });
-        } else {
-          logger.info("Detected LLM model", {
-            model: modelId,
-            maxLen: modelData.max_model_len,
-            root: modelData.root
-          });
-        }
-        return modelId;
-      }
-      if (silent) {
-        logger.debug("LLM server returned empty models list");
-      } else {
-        logger.warn("LLM server returned empty models list");
-      }
-      llmStatus = "disconnected";
-      return null;
-    } catch (error) {
-      const errMsg = error instanceof Error ? error.message : String(error);
-      if (silent) {
-        logger.debug("Could not detect model from LLM server", { error: errMsg, attempt });
-      } else {
-        logger.warn("Could not detect model from LLM server", { error: errMsg, attempt });
-      }
-      if (attempt < retries) {
-        await new Promise((r) => setTimeout(r, 1e3 * attempt));
-        continue;
-      }
-    }
-  }
-  llmStatus = "disconnected";
-  return cached?.model ?? null;
-}
-function getLlmStatus() {
-  return llmStatus;
-}
-function clearModelCache(url) {
-  if (url) {
-    modelCache.delete(getCacheKey(url));
-  } else {
-    modelCache.clear();
-  }
-  llmStatus = "unknown";
-}
-
 export {
-  SessionNotFoundError,
-  ensureVersionPrefix,
-  stripVersionPrefix,
-  buildModelsUrl,
-  createLLMClient,
-  detectModel,
-  getLlmStatus,
-  clearModelCache
+  createLLMClient
 };
-//# sourceMappingURL=chunk-XJ4SUDL7.js.map
+//# sourceMappingURL=chunk-QYP6MOB5.js.map

package/dist/chunk-V4IE7HJY.js

@@ -0,0 +1,175 @@
+// src/server/llm/profiles.ts
+var DEFAULT_PROFILE = {
+  name: "default",
+  temperature: 0.7,
+  topP: 0.9,
+  defaultMaxTokens: 4096,
+  supportsVision: true
+};
+var MOCK_PROFILE = {
+  name: "mock",
+  temperature: 0.7,
+  topP: 0.9,
+  defaultMaxTokens: 1024,
+  supportsVision: false
+};
+var MODEL_PROFILES = [
+  {
+    pattern: "mistral",
+    profile: {
+      name: "Mistral",
+      temperature: 0.7,
+      topP: 0.9,
+      defaultMaxTokens: 16384,
+      supportsVision: false
+    }
+  },
+  {
+    pattern: "qwen3-coder-next",
+    profile: {
+      name: "Qwen3-Coder-Next",
+      // Per Qwen docs: "temperature=1.0, top_p=0.95, top_k=40"
+      temperature: 1,
+      topP: 0.95,
+      topK: 40,
+      // "This model supports only non-thinking mode and does not generate <think></think> blocks"
+      defaultMaxTokens: 16384,
+      supportsVision: false
+    }
+  },
+  {
+    pattern: "qwen3",
+    profile: {
+      name: "Qwen3",
+      temperature: 0.7,
+      topP: 0.9,
+      defaultMaxTokens: 16384,
+      supportsVision: false
+    }
+  },
+  {
+    pattern: "qwen3-vl",
+    profile: {
+      name: "Qwen3-VL",
+      temperature: 0.7,
+      topP: 0.9,
+      defaultMaxTokens: 16384,
+      supportsVision: true
+    }
+  },
+  {
+    pattern: "deepseek",
+    profile: {
+      name: "DeepSeek",
+      temperature: 0.6,
+      topP: 0.95,
+      defaultMaxTokens: 16384,
+      supportsVision: false
+    }
+  },
+  {
+    pattern: "minimax-m2.5",
+    profile: {
+      name: "MiniMax-M2.5",
+      temperature: 1,
+      topP: 0.95,
+      topK: 40,
+      defaultMaxTokens: 16384,
+      supportsVision: false
+    }
+  },
+  {
+    pattern: "minimax-m2.7",
+    profile: {
+      name: "MiniMax-M2.7",
+      temperature: 1,
+      topP: 0.95,
+      topK: 40,
+      defaultMaxTokens: 16384,
+      supportsVision: false
+    }
+  },
+  {
+    pattern: "minimax",
+    profile: {
+      name: "MiniMax",
+      temperature: 1,
+      topP: 0.95,
+      topK: 40,
+      defaultMaxTokens: 16384,
+      supportsVision: false
+    }
+  },
+  {
+    pattern: "minimax-m3",
+    profile: {
+      name: "MiniMax-M3",
+      temperature: 1,
+      topP: 0.95,
+      topK: 40,
+      defaultMaxTokens: 16384,
+      supportsVision: true
+    }
+  },
+  {
+    pattern: "llava",
+    profile: {
+      name: "LLaVA",
+      temperature: 0.7,
+      topP: 0.9,
+      defaultMaxTokens: 16384,
+      supportsVision: true
+    }
+  },
+  {
+    pattern: "llama",
+    profile: {
+      name: "Llama",
+      temperature: 0.7,
+      topP: 0.9,
+      defaultMaxTokens: 16384,
+      supportsVision: false
+    }
+  },
+  {
+    pattern: "claude",
+    profile: {
+      name: "Claude",
+      temperature: 0.7,
+      topP: 0.9,
+      defaultMaxTokens: 16384,
+      supportsVision: false
+    }
+  },
+  {
+    pattern: "gemma-4",
+    profile: {
+      name: "Gemma 4",
+      temperature: 0.7,
+      topP: 0.9,
+      defaultMaxTokens: 16384,
+      supportsVision: true
+    }
+  }
+];
+function getModelProfile(modelName) {
+  const lowerName = modelName.toLowerCase();
+  if (lowerName.includes("mock")) {
+    return MOCK_PROFILE;
+  }
+  for (const { pattern, profile } of MODEL_PROFILES) {
+    if (lowerName.includes(pattern.toLowerCase())) {
+      return profile;
+    }
+  }
+  return DEFAULT_PROFILE;
+}
+function modelSupportsVision(modelName) {
+  return getModelProfile(modelName).supportsVision;
+}
+
+export {
+  getModelProfile,
+  modelSupportsVision
+};
+//# sourceMappingURL=chunk-V4IE7HJY.js.map

package/dist/{chunk-GI24G4OW.js → chunk-VI236SOY.js}

@@ -1,3 +1,10 @@
+import {
+  createContextStateMessage,
+  createSessionStateMessage
+} from "./chunk-F4PMNP7S.js";
+import {
+  getPendingQuestionsForSession
+} from "./chunk-EU3WWTFH.js";
 import {
   getEventStore,
   updateSessionMetadata
@@ -6,13 +13,6 @@ import {
   buildMessagesFromStoredEvents,
   foldPendingConfirmations
 } from "./chunk-6PLAWCHQ.js";
-import {
-  createContextStateMessage,
-  createSessionStateMessage
-} from "./chunk-F4PMNP7S.js";
-import {
-  getPendingQuestionsForSession
-} from "./chunk-EU3WWTFH.js";
 import {
   logger
 } from "./chunk-K44MW7JJ.js";
@@ -58,53 +58,6 @@ Example inputs and outputs:
 - "Add unit tests for the API" \u2192 "Add API unit tests"
 
 User message: {message}`;
-async function generateSessionName(options) {
-  const { userMessage, llmClient, signal, modelSettings, nonThinkingEnabled } = options;
-  try {
-    logger.debug("Generating session name", { messagePreview: userMessage.slice(0, 50) });
-    const prompt = SESSION_NAME_PROMPT.replace("{message}", userMessage);
-    const messages = [
-      {
-        role: "user",
-        content: prompt
-      }
-    ];
-    const timeoutSignal = AbortSignal.timeout(6e4);
-    const composedSignal = signal ? AbortSignal.any([timeoutSignal, signal]) : timeoutSignal;
-    const response = await llmClient.complete({
-      messages,
-      tools: [],
-      signal: composedSignal,
-      // Default to non-thinking (reasoningEffort: 'none') when not explicitly configured
-      // to prevent thinking output in session names. Only skip when user explicitly
-      // set nonThinkingEnabled to false.
-      ...nonThinkingEnabled !== false ? { reasoningEffort: "none" } : {},
-      ...modelSettings ? { modelSettings } : {}
-    });
-    let name = (response.content || response.thinkingContent || "").trim();
-    if (name.length > 50) {
-      name = name.substring(0, 47) + "...";
-    }
-    if (!name || name.length < 3) {
-      return {
-        success: false,
-        error: "Generated name is too short or empty"
-      };
-    }
-    logger.debug("Session name generated successfully", { name });
-    return {
-      success: true,
-      name
-    };
-  } catch (error) {
-    const errorMessage = error instanceof Error ? error.message : "Unknown error generating session name";
-    logger.debug("Session name generation error", { error: errorMessage });
-    return {
-      success: false,
-      error: errorMessage
-    };
-  }
-}
 function needsNameGeneration(sessionTitle, messageCount) {
   if (messageCount > 1) {
     return false;
@@ -145,13 +98,80 @@ function applyGeneratedSessionName(sessionId, name, deps) {
     );
   }
 }
+function resolveSessionProvider(session, providerManager) {
+  const providers = providerManager.getProviders();
+  const effectiveModel = session.providerModel ?? providerManager.getCurrentModel();
+  if (!effectiveModel) return void 0;
+  const provider = session.providerId ? providers.find((p) => p.id === session.providerId) : providers.find((p) => p.isActive);
+  if (!provider) return void 0;
+  return {
+    baseUrl: provider.url,
+    ...provider.apiKey ? { apiKey: provider.apiKey } : {},
+    model: effectiveModel
+  };
+}
+async function generateSessionNameForSession(sessionId, userMessage, deps, signal) {
+  const { sessionManager, providerManager, broadcastForSession, eventStore } = deps;
+  const session = sessionManager.getSession(sessionId);
+  if (!session) return;
+  const messageCount = getSessionMessageCount(sessionId);
+  if (!needsNameGenerationCheck(sessionId, session.metadata.title, messageCount)) return;
+  const providerConfig = resolveSessionProvider(session, providerManager);
+  const modelSettings = providerConfig ? providerManager.getModelSettings(providerConfig.model, "non-thinking") : void 0;
+  const prompt = SESSION_NAME_PROMPT.replace("{message}", userMessage);
+  try {
+    let client;
+    if (deps.getLLMClient && providerConfig) {
+      client = deps.getLLMClient();
+      client.setModel(providerConfig.model);
+    } else if (providerConfig) {
+      const { createLLMClient } = await import("./client-X6BVH4Q4.js");
+      client = createLLMClient({
+        llm: {
+          baseUrl: providerConfig.baseUrl,
+          model: providerConfig.model,
+          ...providerConfig.apiKey ? { apiKey: providerConfig.apiKey } : {}
+        }
+      });
+    } else if (deps.getLLMClient) {
+      client = deps.getLLMClient();
+    } else {
+      logger.debug("Session name generation skipped: no LLM client available", { sessionId });
+      return;
+    }
+    const timeoutSignal = AbortSignal.timeout(12e4);
+    const composedSignal = signal ? AbortSignal.any([timeoutSignal, signal]) : timeoutSignal;
+    const response = await client.complete({
+      messages: [{ role: "user", content: prompt }],
+      tools: [],
+      ...modelSettings ? { modelSettings } : {},
+      signal: composedSignal,
+      skipClientReasoningEffort: true
+    });
+    let name = (response.content || response.thinkingContent || "").trim();
+    if (name.length > 50) {
+      name = name.substring(0, 47) + "...";
+    }
+    if (name.length >= 3) {
+      applyGeneratedSessionName(sessionId, name, {
+        sessionManager,
+        eventStore,
+        broadcastForSession
+      });
+    } else {
+      logger.debug("Session name too short", { sessionId, name });
+    }
+  } catch (error) {
+    logger.error("Session name generation error", {
+      sessionId,
+      error: error instanceof Error ? error.message : String(error)
+    });
+  }
+}
 
 export {
-  getSessionMessageCount,
   finalizeTurnCompletion,
   buildRunChatTurnParams,
-  generateSessionName,
-  needsNameGenerationCheck,
-  applyGeneratedSessionName
+  generateSessionNameForSession
 };
-//# sourceMappingURL=chunk-GI24G4OW.js.map
+//# sourceMappingURL=chunk-VI236SOY.js.map

package/dist/{chunk-CDIYCGCO.js → chunk-WEXW7ZXJ.js}

@@ -1,7 +1,7 @@
 import {
   COMPACTION_PROMPT,
   createMessageStartEvent
-} from "./chunk-YUHODMKY.js";
+} from "./chunk-YGSBVKFU.js";
 import {
   getCurrentWindowMessageOptions
 } from "./chunk-YBWY4DKY.js";
@@ -31,4 +31,4 @@ export {
   shouldCompact,
   getCompactionTarget
 };
-//# sourceMappingURL=chunk-CDIYCGCO.js.map
+//# sourceMappingURL=chunk-WEXW7ZXJ.js.map