lynkr 7.2.5 → 8.0.1

package/README.md

@@ -50,7 +50,7 @@ Lynkr is a **self-hosted proxy server** that unlocks Claude Code CLI , Cursor ID
 npm install -g pino-pretty 
 npm install -g lynkr
 
-lynk start
+lynkr start
 ```
 
 **Option 2: Git Clone**
@@ -238,7 +238,7 @@ Lynkr supports [ClawdBot](https://github.com/openclaw/openclaw) via its OpenAI-c
 
 ### Getting Started
 - 📦 **[Installation Guide](documentation/installation.md)** - Detailed installation for all methods
-- ⚙️ **[Provider Configuration](documentation/providers.md)** - Complete setup for all 9+ providers
+- ⚙️ **[Provider Configuration](documentation/providers.md)** - Complete setup for all 12+ providers
 - 🎯 **[Quick Start Examples](documentation/installation.md#quick-start-examples)** - Copy-paste configs
 
 ### IDE & CLI Integration
@@ -277,7 +277,7 @@ Lynkr supports [ClawdBot](https://github.com/openclaw/openclaw) via its OpenAI-c
 
 ## Key Features Highlights
 
-- ✅ **Multi-Provider Support** - 9+ providers including local (Ollama, llama.cpp) and cloud (Bedrock, Databricks, OpenRouter)
+- ✅ **Multi-Provider Support** - 12+ providers including local (Ollama, llama.cpp) and cloud (Bedrock, Databricks, OpenRouter, Moonshot AI)
 - ✅ **60-80% Cost Reduction** - Token optimization with smart tool selection, prompt caching, memory deduplication
 - ✅ **100% Local Option** - Run completely offline with Ollama/llama.cpp (zero cloud dependencies)
 - ✅ **OpenAI Compatible** - Works with Cursor IDE, Continue.dev, and any OpenAI-compatible client

package/config/model-tiers.json

@@ -0,0 +1,89 @@
+{
+  "tiers": {
+    "SIMPLE": {
+      "description": "Greetings, simple Q&A, confirmations, basic lookups",
+      "range": [0, 25],
+      "priority": 1,
+      "preferred": {
+        "ollama": ["llama3.2", "gemma2", "phi3", "qwen2.5:7b", "mistral"],
+        "llamacpp": ["default"],
+        "lmstudio": ["default"],
+        "openai": ["gpt-4o-mini", "gpt-3.5-turbo"],
+        "azure-openai": ["gpt-4o-mini", "gpt-35-turbo"],
+        "anthropic": ["claude-3-haiku-20240307", "claude-3-5-haiku-20241022"],
+        "bedrock": ["anthropic.claude-3-haiku-20240307-v1:0", "amazon.nova-lite-v1:0"],
+        "databricks": ["databricks-claude-haiku-4-5", "databricks-gpt-5-nano"],
+        "google": ["gemini-2.0-flash", "gemini-1.5-flash"],
+        "openrouter": ["google/gemini-flash-1.5", "deepseek/deepseek-chat"],
+        "zai": ["GLM-4-Flash"],
+        "moonshot": ["kimi-k2-turbo-preview"]
+      }
+    },
+    "MEDIUM": {
+      "description": "Code reading, simple edits, research, documentation",
+      "range": [26, 50],
+      "priority": 2,
+      "preferred": {
+        "ollama": ["qwen2.5:32b", "deepseek-coder:33b", "codellama:34b"],
+        "llamacpp": ["default"],
+        "lmstudio": ["default"],
+        "openai": ["gpt-4o", "gpt-4-turbo"],
+        "azure-openai": ["gpt-4o", "gpt-4"],
+        "anthropic": ["claude-sonnet-4-20250514", "claude-3-5-sonnet-20241022"],
+        "bedrock": ["anthropic.claude-3-5-sonnet-20241022-v2:0", "amazon.nova-pro-v1:0"],
+        "databricks": ["databricks-claude-sonnet-4-5", "databricks-gpt-5-1"],
+        "google": ["gemini-1.5-pro", "gemini-2.0-pro"],
+        "openrouter": ["anthropic/claude-3.5-sonnet", "openai/gpt-4o"],
+        "zai": ["GLM-4.7"],
+        "moonshot": ["kimi-k2-turbo-preview"]
+      }
+    },
+    "COMPLEX": {
+      "description": "Multi-file changes, debugging, architecture, refactoring",
+      "range": [51, 75],
+      "priority": 3,
+      "preferred": {
+        "ollama": ["qwen2.5:72b", "llama3.1:70b", "deepseek-coder-v2:236b"],
+        "openai": ["o1-mini", "o3-mini", "gpt-4o"],
+        "azure-openai": ["o1-mini", "gpt-4o"],
+        "anthropic": ["claude-sonnet-4-20250514", "claude-3-5-sonnet-20241022"],
+        "bedrock": ["anthropic.claude-3-5-sonnet-20241022-v2:0"],
+        "databricks": ["databricks-claude-sonnet-4-5", "databricks-gpt-5-1-codex-max"],
+        "google": ["gemini-2.5-pro", "gemini-1.5-pro"],
+        "openrouter": ["anthropic/claude-3.5-sonnet", "meta-llama/llama-3.1-405b"],
+        "zai": ["GLM-4.7"],
+        "moonshot": ["kimi-k2-turbo-preview"]
+      }
+    },
+    "REASONING": {
+      "description": "Complex analysis, security audits, novel problems, deep thinking",
+      "range": [76, 100],
+      "priority": 4,
+      "preferred": {
+        "openai": ["o1", "o1-pro", "o3"],
+        "azure-openai": ["o1", "o1-pro"],
+        "anthropic": ["claude-opus-4-20250514", "claude-3-opus-20240229"],
+        "bedrock": ["anthropic.claude-3-opus-20240229-v1:0"],
+        "databricks": ["databricks-claude-opus-4-6", "databricks-claude-opus-4-5", "databricks-gpt-5-2"],
+        "google": ["gemini-2.5-pro"],
+        "openrouter": ["anthropic/claude-3-opus", "deepseek/deepseek-reasoner", "openai/o1"],
+        "deepseek": ["deepseek-reasoner", "deepseek-r1"],
+        "moonshot": ["kimi-k2-thinking", "kimi-k2-turbo-preview"]
+      }
+    }
+  },
+  "localProviders": {
+    "ollama": { "free": true, "defaultTier": "SIMPLE" },
+    "llamacpp": { "free": true, "defaultTier": "SIMPLE" },
+    "lmstudio": { "free": true, "defaultTier": "SIMPLE" }
+  },
+  "providerAliases": {
+    "azure": "azure-openai",
+    "aws": "bedrock",
+    "amazon": "bedrock",
+    "claude": "anthropic",
+    "gemini": "google",
+    "vertex": "google",
+    "kimi": "moonshot"
+  }
+}

package/install.sh

@@ -134,10 +134,15 @@ MODEL_PROVIDER=ollama
 PORT=8080
 
 # Ollama Configuration (default for local development)
-PREFER_OLLAMA=true
 OLLAMA_MODEL=qwen2.5-coder:7b
 OLLAMA_ENDPOINT=http://localhost:11434
 
+# Tier-based routing (uncomment and configure to enable)
+# TIER_SIMPLE=ollama:qwen2.5-coder:7b
+# TIER_MEDIUM=ollama:qwen2.5-coder:7b
+# TIER_COMPLEX=ollama:qwen2.5-coder:7b
+# TIER_REASONING=ollama:qwen2.5-coder:7b
+
 # Long-Term Memory System (Titans-Inspired) - Enabled by default
 MEMORY_ENABLED=true
 MEMORY_RETRIEVAL_LIMIT=5

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "lynkr",
-  "version": "7.2.5",
+  "version": "8.0.1",
   "description": "Self-hosted Claude Code & Cursor proxy with Databricks,AWS BedRock,Azure  adapters, openrouter, Ollama,llamacpp,LM Studio, workspace tooling, and MCP integration.",
   "main": "index.js",
   "bin": {
@@ -14,7 +14,7 @@
     "dev": "nodemon index.js",
     "lint": "eslint src index.js",
     "test": "npm run test:unit && npm run test:performance",
-    "test:unit": "DATABRICKS_API_KEY=test-key DATABRICKS_API_BASE=http://test.com node --test test/routing.test.js test/hybrid-routing-integration.test.js test/web-tools.test.js test/passthrough-mode.test.js test/openrouter-error-resilience.test.js test/format-conversion.test.js test/azure-openai-config.test.js test/azure-openai-format-conversion.test.js test/azure-openai-routing.test.js test/azure-openai-streaming.test.js test/azure-openai-error-resilience.test.js test/azure-openai-integration.test.js test/openai-integration.test.js test/llamacpp-integration.test.js test/memory/store.test.js test/memory/surprise.test.js test/memory/extractor.test.js test/memory/search.test.js test/memory/retriever.test.js",
+    "test:unit": "DATABRICKS_API_KEY=test-key DATABRICKS_API_BASE=http://test.com node --test test/routing.test.js test/hybrid-routing-integration.test.js test/web-tools.test.js test/passthrough-mode.test.js test/openrouter-error-resilience.test.js test/format-conversion.test.js test/azure-openai-config.test.js test/azure-openai-format-conversion.test.js test/azure-openai-routing.test.js test/azure-openai-streaming.test.js test/azure-openai-error-resilience.test.js test/azure-openai-integration.test.js test/openai-integration.test.js test/toon-compression.test.js test/llamacpp-integration.test.js test/memory/store.test.js test/memory/surprise.test.js test/memory/extractor.test.js test/memory/search.test.js test/memory/retriever.test.js",
     "test:memory": "DATABRICKS_API_KEY=test-key DATABRICKS_API_BASE=http://test.com node --test test/memory/store.test.js test/memory/surprise.test.js test/memory/extractor.test.js test/memory/search.test.js test/memory/retriever.test.js",
     "test:new-features": "DATABRICKS_API_KEY=test-key DATABRICKS_API_BASE=http://test.com node --test test/passthrough-mode.test.js test/openrouter-error-resilience.test.js test/format-conversion.test.js",
     "test:performance": "DATABRICKS_API_KEY=test-key DATABRICKS_API_BASE=http://test.com node test/hybrid-routing-performance.test.js && DATABRICKS_API_KEY=test-key DATABRICKS_API_BASE=http://test.com node test/performance-tests.js",
@@ -47,6 +47,7 @@
     "@azure/openai": "^2.0.0",
     "@babel/parser": "^7.29.0",
     "@babel/traverse": "^7.29.0",
+    "@toon-format/toon": "^2.1.0",
     "compression": "^1.7.4",
     "diff": "^5.2.0",
     "dotenv": "^16.4.5",
@@ -57,6 +58,7 @@
     "openai": "^6.14.0",
     "pino": "^8.17.2",
     "pino-http": "^8.6.0",
+    "pino-roll": "^4.0.0",
     "undici": "^6.22.0"
   },
   "optionalDependencies": {

package/scripts/setup.js

@@ -251,7 +251,6 @@ async function createEnvFile() {
     if (ollamaOnly) {
       let envContent = fs.readFileSync(envPath, "utf-8");
       envContent = envContent.replace(/^# MODEL_PROVIDER=databricks/m, "MODEL_PROVIDER=ollama");
-      envContent = envContent.replace(/^PREFER_OLLAMA=true/m, "# PREFER_OLLAMA=true  # Not needed when MODEL_PROVIDER=ollama");
       envContent = envContent.replace(/^FALLBACK_ENABLED=true/m, "FALLBACK_ENABLED=false");
       fs.writeFileSync(envPath, envContent);
     }

package/src/agents/executor.js

@@ -162,14 +162,22 @@ class SubagentExecutor {
       payload.tools = filteredTools;
     }
 
-    // Determine provider based on model - subagents should use the specified model
+    // Determine provider based on model family.
+    // Subagents should use the currently configured MODEL_PROVIDER and avoid
+    // hard-fallbacks to Azure when Azure is not selected/configured.
     let forceProvider = null;
-    if (payload.model?.includes('claude') || payload.model?.includes('sonnet') || payload.model?.includes('haiku') || payload.model?.includes('opus')) {
-      // Route Claude models to the configured Claude provider (azure-openai, databricks, etc.)
+    const modelLower = String(payload.model || "").toLowerCase();
+    const isClaudeFamilyModel =
+      modelLower.includes("claude") ||
+      modelLower.includes("sonnet") ||
+      modelLower.includes("haiku") ||
+      modelLower.includes("opus");
+    const isGptFamilyModel = modelLower.includes("gpt");
+
+    if (isClaudeFamilyModel || isGptFamilyModel) {
       const config = require('../config');
-      forceProvider = config.modelProvider?.provider || 'azure-openai';
-    } else if (payload.model?.includes('gpt')) {
-      forceProvider = 'azure-openai';
+      // `type` is the canonical key; `provider` kept as legacy fallback.
+      forceProvider = config.modelProvider?.type || config.modelProvider?.provider || null;
     }
 
     logger.debug({

package/src/api/middleware/session.js

@@ -45,8 +45,21 @@ function sessionMiddleware(req, res, next) {
     // Add sessionId to logger context for this request
     req.log = logger.child({ sessionId });
 
-    const session = getOrCreateSession(sessionId);
-    req.session = session;
+    // Skip DB persistence for auto-generated (ephemeral) session IDs.
+    // These are created when the client doesn't send a session header,
+    // so storing them just bloats the DB with throwaway records.
+    if (req.generatedSessionId) {
+      req.session = {
+        id: sessionId,
+        createdAt: Date.now(),
+        updatedAt: Date.now(),
+        metadata: {},
+        history: [],
+        _ephemeral: true,
+      };
+    } else {
+      req.session = getOrCreateSession(sessionId);
+    }
     return next();
   } catch (err) {
     return next(err);

package/src/api/openai-router.js

@@ -21,9 +21,9 @@ const orchestrator = require("../orchestrator");
 const { getSession } = require("../sessions");
 const {
   convertOpenAIToAnthropic,
-  convertAnthropicToOpenAI,
-  convertAnthropicStreamChunkToOpenAI
+  convertAnthropicToOpenAI
 } = require("../clients/openai-format");
+const { IDE_SAFE_TOOLS } = require("../clients/standard-tools");
 
 const router = express.Router();
 
@@ -60,13 +60,14 @@ function detectClient(headers) {
  */
 const CLIENT_TOOL_MAPPINGS = {
   // ============== CODEX CLI ==============
-  // Tools: shell_command, read_file, write_file, apply_patch, glob_file_search, rg, list_dir
+  // Confirmed tools: shell, apply_patch, read_file, write_file, list_dir, glob_file_search,
+  //   rg, web_search, update_plan, view_image, memory
+  // NOT supported: spawn_agent/spawn_thread (Task has no Codex equivalent)
   codex: {
     "Bash": {
-      name: "shell_command",
+      name: "shell",
       mapArgs: (a) => ({
-        command: a.command || "",
-        workdir: a.cwd || a.working_directory
+        command: ["bash", "-c", a.command || ""]
       })
     },
     "Read": {
@@ -113,6 +114,25 @@ const CLIENT_TOOL_MAPPINGS = {
       mapArgs: (a) => ({
         path: a.path || a.directory
       })
+    },
+    "TodoWrite": {
+      name: "update_plan",
+      mapArgs: (a) => ({
+        todos: a.todos || []
+      })
+    },
+    "WebSearch": {
+      name: "web_search",
+      mapArgs: (a) => ({
+        query: a.query || ""
+      })
+    },
+    "WebAgent": {
+      name: "web_agent",
+      mapArgs: (a) => ({
+        url: a.url || "",
+        goal: a.goal || ""
+      })
     }
   },
 
@@ -168,6 +188,13 @@ const CLIENT_TOOL_MAPPINGS = {
         path: a.path || a.directory || ".",
         recursive: false
       })
+    },
+    "WebAgent": {
+      name: "web_agent",
+      mapArgs: (a) => ({
+        url: a.url || "",
+        goal: a.goal || ""
+      })
     }
   },
 
@@ -224,6 +251,13 @@ const CLIENT_TOOL_MAPPINGS = {
         path: a.path || a.directory || ".",
         recursive: false
       })
+    },
+    "WebAgent": {
+      name: "web_agent",
+      mapArgs: (a) => ({
+        url: a.url || "",
+        goal: a.goal || ""
+      })
     }
   },
 
@@ -274,6 +308,13 @@ const CLIENT_TOOL_MAPPINGS = {
       mapArgs: (a) => ({
         filepath: a.path || a.directory || "."
       })
+    },
+    "WebAgent": {
+      name: "web_agent",
+      mapArgs: (a) => ({
+        url: a.url || "",
+        goal: a.goal || ""
+      })
     }
   }
 };
@@ -321,14 +362,7 @@ function mapToolForClient(toolName, argsJson, clientType) {
   };
 }
 
-/**
- * Check if client is a known AI coding tool that needs tool mapping
- * @param {Object} headers - Request headers
- * @returns {boolean}
- */
-function isKnownClient(headers) {
-  return detectClient(headers) !== "unknown";
-}
+
 
 /**
  * POST /v1/chat/completions
@@ -374,7 +408,7 @@ router.post("/chat/completions", async (req, res) => {
         : JSON.stringify(m.content).substring(0, 200)
     }));
 
-    logger.info({
+    logger.debug({
       endpoint: "/v1/chat/completions",
       model: req.body.model,
       messageCount: req.body.messages?.length,
@@ -393,6 +427,32 @@ router.post("/chat/completions", async (req, res) => {
     // Convert OpenAI request to Anthropic format
     const anthropicRequest = convertOpenAIToAnthropic(req.body);
 
+    // Inject tools if client didn't send any.
+    // Two-layer filtering:
+    //   1. IDE_SAFE_TOOLS = STANDARD_TOOLS minus AskUserQuestion (can't work through proxy)
+    //   2. For known clients (codex, cline, etc.), further filter to only tools
+    //      that have a mapping in CLIENT_TOOL_MAPPINGS — this ensures clients like
+    //      Codex don't see tools they can't handle (Task, WebFetch, NotebookEdit)
+    //      while Claude Code (unknown client) gets the full IDE_SAFE_TOOLS set.
+    // Skip injection if client explicitly opted out (tool_choice: "none" or empty tools array).
+    const clientType = detectClient(req.headers);
+    const clientExplicitlyDisabledTools = req.body.tool_choice === "none" || Array.isArray(req.body.tools);
+    if (!clientExplicitlyDisabledTools && (!anthropicRequest.tools || anthropicRequest.tools.length === 0)) {
+      const clientMappings = CLIENT_TOOL_MAPPINGS[clientType];
+      const clientTools = clientMappings
+        ? IDE_SAFE_TOOLS.filter(t => clientMappings[t.name])
+        : IDE_SAFE_TOOLS;
+      anthropicRequest.tools = clientTools;
+      logger.debug({
+        clientType,
+        injectedToolCount: clientTools.length,
+        injectedToolNames: clientTools.map(t => t.name),
+        reason: clientMappings
+          ? `Known client '${clientType}' — filtered to mapped tools only`
+          : "Unknown client — injecting full IDE_SAFE_TOOLS"
+      }, "=== INJECTING TOOLS ===");
+    }
+
     // Get or create session
     const session = getSession(sessionId);
 
@@ -420,7 +480,7 @@ router.post("/chat/completions", async (req, res) => {
         });
 
         // Check if we have a valid response body
-        logger.info({
+        logger.debug({
           hasResult: !!result,
           resultKeys: result ? Object.keys(result) : null,
           hasBody: result && !!result.body,
@@ -442,7 +502,7 @@ router.post("/chat/completions", async (req, res) => {
         const openaiResponse = convertAnthropicToOpenAI(result.body, req.body.model);
 
         // Debug: Log what we're about to stream
-        logger.info({
+        logger.debug({
           openaiResponseId: openaiResponse.id,
           messageContent: openaiResponse.choices[0]?.message?.content?.substring(0, 100),
           contentLength: openaiResponse.choices[0]?.message?.content?.length || 0,
@@ -454,7 +514,25 @@ router.post("/chat/completions", async (req, res) => {
 
         // Simulate streaming by sending the complete response as chunks
         const content = openaiResponse.choices[0].message.content || "";
-        const toolCalls = openaiResponse.choices[0].message.tool_calls;
+        let toolCalls = openaiResponse.choices[0].message.tool_calls;
+
+        // Map tool names for known IDE clients
+        if (clientType !== "unknown" && toolCalls && toolCalls.length > 0) {
+          toolCalls = toolCalls.map(tc => {
+            const mapped = mapToolForClient(tc.function?.name || "", tc.function?.arguments || "{}", clientType);
+            return {
+              ...tc,
+              function: {
+                name: mapped.name,
+                arguments: mapped.arguments
+              }
+            };
+          });
+          logger.debug({
+            mappedTools: toolCalls.map(t => t.function?.name),
+            clientType
+          }, "Tool names mapped for streaming chat/completions");
+        }
 
         // Send start chunk with role
         const startChunk = {
@@ -493,7 +571,7 @@ router.post("/chat/completions", async (req, res) => {
             }]
           };
           const contentWriteOk = res.write(`data: ${JSON.stringify(contentChunk)}\n\n`);
-          logger.info({ contentPreview: content.substring(0, 50), writeOk: contentWriteOk }, "Sent content chunk");
+          logger.debug({ contentPreview: content.substring(0, 50), writeOk: contentWriteOk }, "Sent content chunk");
         }
 
         // Send tool calls if present
@@ -545,7 +623,7 @@ router.post("/chat/completions", async (req, res) => {
         res.write("data: [DONE]\n\n");
 
         // Ensure data is flushed before ending
-        logger.info({ contentLength: content.length, contentPreview: content.substring(0, 50) }, "=== SSE STREAM COMPLETE ===");
+        logger.debug({ contentLength: content.length, contentPreview: content.substring(0, 50) }, "=== SSE STREAM COMPLETE ===");
         res.end();
 
         logger.info({
@@ -558,10 +636,7 @@ router.post("/chat/completions", async (req, res) => {
       } catch (streamError) {
         logger.error({
           error: streamError.message,
-          stack: streamError.stack,
-          resultWasNull: !result,
-          resultBodyWasNull: result && !result.body,
-          resultKeys: result ? Object.keys(result) : null
+          stack: streamError.stack
         }, "=== STREAMING ERROR ===");
 
         // Send error in OpenAI streaming format
@@ -606,6 +681,24 @@ router.post("/chat/completions", async (req, res) => {
       // Convert Anthropic response to OpenAI format
       const openaiResponse = convertAnthropicToOpenAI(result.body, req.body.model);
 
+      // Map tool names for known IDE clients
+      if (clientType !== "unknown" && openaiResponse.choices?.[0]?.message?.tool_calls?.length > 0) {
+        openaiResponse.choices[0].message.tool_calls = openaiResponse.choices[0].message.tool_calls.map(tc => {
+          const mapped = mapToolForClient(tc.function?.name || "", tc.function?.arguments || "{}", clientType);
+          return {
+            ...tc,
+            function: {
+              name: mapped.name,
+              arguments: mapped.arguments
+            }
+          };
+        });
+        logger.debug({
+          mappedTools: openaiResponse.choices[0].message.tool_calls.map(t => t.function?.name),
+          clientType
+        }, "Tool names mapped for non-streaming chat/completions");
+      }
+
       logger.info({
         duration: Date.now() - startTime,
         mode: "non-streaming",
@@ -767,6 +860,18 @@ function getConfiguredProviders() {
     });
   }
 
+  // Check Moonshot AI (Kimi)
+  if (config.moonshot?.apiKey) {
+    providers.push({
+      name: "moonshot",
+      type: "moonshot-ai",
+      models: [
+        config.moonshot.model || "kimi-k2-turbo-preview",
+        "kimi-k2-turbo-preview"
+      ]
+    });
+  }
+
   // Check Vertex AI (Google Cloud)
   if (config.vertex?.projectId) {
     providers.push({
@@ -1013,7 +1118,7 @@ function determineEmbeddingProvider(requestedModel = null) {
 async function generateOllamaEmbeddings(inputs, embeddingConfig) {
   const { model, endpoint } = embeddingConfig;
 
-  logger.info({
+  logger.debug({
     model,
     endpoint,
     inputCount: inputs.length
@@ -1079,7 +1184,7 @@ async function generateOllamaEmbeddings(inputs, embeddingConfig) {
 async function generateLlamaCppEmbeddings(inputs, embeddingConfig) {
   const { model, endpoint } = embeddingConfig;
 
-  logger.info({
+  logger.debug({
     model,
     endpoint,
     inputCount: inputs.length
@@ -1147,7 +1252,7 @@ async function generateLlamaCppEmbeddings(inputs, embeddingConfig) {
 async function generateOpenRouterEmbeddings(inputs, embeddingConfig) {
   const { model, apiKey, endpoint } = embeddingConfig;
 
-  logger.info({
+  logger.debug({
     model,
     inputCount: inputs.length
   }, "Generating embeddings with OpenRouter");
@@ -1181,7 +1286,7 @@ async function generateOpenRouterEmbeddings(inputs, embeddingConfig) {
 async function generateOpenAIEmbeddings(inputs, embeddingConfig) {
   const { model, apiKey, endpoint } = embeddingConfig;
 
-  logger.info({
+  logger.debug({
     model,
     inputCount: inputs.length
   }, "Generating embeddings with OpenAI");
@@ -1233,7 +1338,7 @@ router.post("/embeddings", async (req, res) => {
     // Convert input to array if string
     const inputs = Array.isArray(input) ? input : [input];
 
-    logger.info({
+    logger.debug({
       endpoint: "/v1/embeddings",
       model: model || "auto-detect",
       inputCount: inputs.length,
@@ -1335,7 +1440,7 @@ router.post("/responses", async (req, res) => {
     const { convertResponsesToChat, convertChatToResponses } = require("../clients/responses-format");
 
     // Comprehensive debug logging
-    logger.info({
+    logger.debug({
       endpoint: "/v1/responses",
       inputType: typeof req.body.input,
       inputIsArray: Array.isArray(req.body.input),
@@ -1354,7 +1459,7 @@ router.post("/responses", async (req, res) => {
     // Convert Responses API to Chat Completions format
     const chatRequest = convertResponsesToChat(req.body);
 
-    logger.info({
+    logger.debug({
       chatRequestMessageCount: chatRequest.messages?.length,
       chatRequestMessages: chatRequest.messages?.map(m => ({
         role: m.role,
@@ -1366,7 +1471,7 @@ router.post("/responses", async (req, res) => {
     // Convert to Anthropic format
     const anthropicRequest = convertOpenAIToAnthropic(chatRequest);
 
-    logger.info({
+    logger.debug({
       anthropicMessageCount: anthropicRequest.messages?.length,
       anthropicMessages: anthropicRequest.messages?.map(m => ({
         role: m.role,
@@ -1374,6 +1479,26 @@ router.post("/responses", async (req, res) => {
       }))
     }, "After Chat→Anthropic conversion");
 
+    // Inject tools if client didn't send any (same two-layer filtering as chat/completions).
+    // Skip injection if client explicitly opted out (tool_choice: "none" or empty tools array).
+    const clientType = detectClient(req.headers);
+    const clientExplicitlyDisabledTools = req.body.tool_choice === "none" || Array.isArray(req.body.tools);
+    if (!clientExplicitlyDisabledTools && (!anthropicRequest.tools || anthropicRequest.tools.length === 0)) {
+      const clientMappings = CLIENT_TOOL_MAPPINGS[clientType];
+      const clientTools = clientMappings
+        ? IDE_SAFE_TOOLS.filter(t => clientMappings[t.name])
+        : IDE_SAFE_TOOLS;
+      anthropicRequest.tools = clientTools;
+      logger.debug({
+        clientType,
+        injectedToolCount: clientTools.length,
+        injectedToolNames: clientTools.map(t => t.name),
+        reason: clientMappings
+          ? `Known client '${clientType}' — filtered to mapped tools only`
+          : "Unknown client — injecting full IDE_SAFE_TOOLS"
+      }, "=== INJECTING TOOLS (responses) ===");
+    }
+
     // Get session
     const session = getSession(sessionId);
 
@@ -1400,7 +1525,7 @@ router.post("/responses", async (req, res) => {
         });
 
         // Debug: Log what orchestrator returned
-        logger.info({
+        logger.debug({
           hasResult: !!result,
           hasBody: !!result?.body,
           bodyKeys: result?.body ? Object.keys(result.body) : null,
@@ -1412,7 +1537,7 @@ router.post("/responses", async (req, res) => {
         // Convert back: Anthropic → OpenAI → Responses
         const chatResponse = convertAnthropicToOpenAI(result.body, req.body.model);
 
-        logger.info({
+        logger.debug({
           chatContent: chatResponse.choices?.[0]?.message?.content?.substring(0, 200),
           chatContentLength: chatResponse.choices?.[0]?.message?.content?.length || 0,
           hasToolCalls: !!chatResponse.choices?.[0]?.message?.tool_calls,
@@ -1433,7 +1558,7 @@ router.post("/responses", async (req, res) => {
         // Check if client is a known AI coding tool and map tool names accordingly
         const clientType = detectClient(req.headers);
         if (clientType !== "unknown" && toolCalls.length > 0) {
-          logger.info({
+          logger.debug({
             originalTools: toolCalls.map(t => t.function?.name),
             clientType,
             userAgent: req.headers["user-agent"]
@@ -1451,12 +1576,12 @@ router.post("/responses", async (req, res) => {
             };
           });
 
-          logger.info({
+          logger.debug({
             mappedTools: toolCalls.map(t => t.function?.name)
           }, `Tool names mapped for ${clientType}`);
         }
 
-        logger.info({
+        logger.debug({
           content: content.substring(0, 100),
           contentLength: content.length,
           toolCallCount: toolCalls.length,

package/src/api/providers-handler.js

@@ -179,6 +179,20 @@ function getConfiguredProviders() {
     });
   }
 
+  // Check Moonshot AI (Kimi)
+  if (config.moonshot?.apiKey) {
+    providers.push({
+      name: "moonshot",
+      type: "moonshot-ai",
+      baseUrl: config.moonshot.endpoint || "https://api.moonshot.ai/v1",
+      enabled: true,
+      models: [
+        { id: config.moonshot.model || "kimi-k2-turbo-preview", name: "Configured Model" },
+        { id: "kimi-k2-turbo-preview", name: "Kimi K2 Turbo Preview" },
+      ]
+    });
+  }
+
   // Check Vertex AI (Google Cloud)
   if (config.vertex?.projectId) {
     const region = config.vertex.region || "us-east5";
@@ -369,7 +383,7 @@ router.get("/config", (req, res) => {
       model_provider: config.modelProvider?.type || "databricks",
       fallback_provider: config.modelProvider?.fallbackProvider || null,
       fallback_enabled: config.modelProvider?.fallbackEnabled || false,
-      prefer_ollama: config.modelProvider?.preferOllama || false,
+      tier_routing_enabled: config.modelTiers?.enabled || false,
       tool_execution_mode: config.toolExecutionMode || "server",
       configured_providers: providers.map(p => p.name),
       memory_enabled: config.memory?.enabled || false,