copilot-custom-endpoint 1.3.9 → 1.3.11

package/docs/models/kimi.md

@@ -17,7 +17,10 @@
 
 ## Quick Start
 
-1. **Start the proxy:** `npm run proxy:kimi`
+1. **Start the proxy** — choose one:
+   - `npm run proxy:kimi` (from the repo root)
+   - `npx copilot-custom-endpoint kimi` (standalone, no clone needed)
+   - `npx copilot-custom-endpoint` (also starts the Qwen proxy concurrently)
 2. **Edit `chatLanguageModels.json`** — add the Kimi block from [Setup](#setup) below.
 3. **Set your Moonshot API key** via the Command Palette → **Chat: Manage Language Models**.
 4. **Restart VS Code** and pick "Kimi K2.6" in the chat picker.

package/docs/models/qwen.md

@@ -11,7 +11,7 @@
 | Tool calling                    | ✅ Yes                                                                    |
 | Context                         | 1M                                                                        |
 | Required `requestBody` (direct) | `enable_thinking: false`                                                  |
-| Required `requestBody` (proxy)  | none — proxy injects based on `tools` presence                            |
+| Required `requestBody` (proxy)  | none — proxy injects based on tool activity in the conversation           |
 | Endpoint                        | `https://dashscope-intl.aliyuncs.com/compatible-mode/v1/chat/completions` |
 | Proxy endpoint                  | `http://127.0.0.1:3458/v1/chat/completions`                               |
 
@@ -32,7 +32,10 @@
 
 ## Quick Start — With Proxy (Dynamic Thinking)
 
-1. **Start the proxy:** `npm run proxy:qwen`.
+1. **Start the proxy** — choose one:
+   - `npm run proxy:qwen` (from the repo root)
+   - `npx copilot-custom-endpoint qwen` (standalone, no clone needed)
+   - `npx copilot-custom-endpoint` (also starts the Kimi proxy concurrently)
 2. **Edit `chatLanguageModels.json`** — use the proxy-path block from [Setup § Proxy](#proxy-path) below.
 3. **Set your DashScope API key** via the Language Models UI.
 4. **Restart VS Code.** Reasoning will be visible in plain chat and suppressed on tool turns.
@@ -162,12 +165,16 @@ All can be set in a `.env` file at the repo root (both proxies `import 'dotenv/c
 
 #### Proxy request rewriting rules
 
-| Condition                                | Action                                                      |
-| ---------------------------------------- | ----------------------------------------------------------- |
-| `body.tools` is a non-empty array        | Set `body.enable_thinking = false`                          |
-| `body.tools` is missing, empty, or falsy | Delete `body.enable_thinking` (let model default to `true`) |
+The proxy detects active tool use by examining the conversation state, not just the `tools` array:
+
+| Condition                                                                                             | Action                                                      |
+| ----------------------------------------------------------------------------------------------------- | ----------------------------------------------------------- |
+| A `"tool"`-role message exists in the conversation **or** `tool_choice` is set to a non-default value | Set `body.enable_thinking = false`                          |
+| No tool-role messages and no non-default `tool_choice` (plain chat)                                   | Delete `body.enable_thinking` (let model default to `true`) |
 
 > **Why delete rather than set `true`?** Omitting the key lets Qwen use its built-in default (`true`). Deletion is closer to "don't interfere."
+>
+> **Why not check `body.tools`?** The proxy checks for tool _activity_ — tool results in the message history or an explicit `tool_choice` directive — rather than the mere presence of a tools array. This correctly handles tool-enabled conversations even when the client sends `tools` in an earlier request but omits it from subsequent turns.
 
 ### API key
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "copilot-custom-endpoint",
-  "version": "1.3.9",
+  "version": "1.3.11",
   "description": "Local proxies for VS Code Copilot custom endpoints — Kimi K2 & Qwen 3.x",
   "license": "MIT",
   "type": "module",
@@ -55,4 +55,4 @@
   "dependencies": {
     "dotenv": "^17.4.2"
   }
-}
+}

package/proxy/kimi-proxy.mjs

@@ -104,7 +104,18 @@ function rewriteKimi(payload) {
   const incomingTemperature = payload.temperature
   const incomingTopP = payload.top_p
   const incomingThinkingType = payload?.thinking?.type
-  const hasTools = Array.isArray(payload.tools) && payload.tools.length > 0
+
+  // Determine if a tool is actually being invoked:
+  // - tool_choice is set and not "none"
+  // - OR there is a "tool" role message in the conversation
+  const messages = Array.isArray(payload.messages) ? payload.messages : []
+  const hasToolRole = messages.some((message) => message?.role === 'tool')
+  const toolChoice = payload.tool_choice
+  const hasActiveToolCall =
+    hasToolRole ||
+    (toolChoice !== undefined && toolChoice !== 'none' && toolChoice !== null)
+  const hasTools = hasActiveToolCall
+
   const useNonThinkingMode = disableThinkingWithTools && hasTools
   const rewrittenTemperature = useNonThinkingMode
     ? forcedNonThinkingTemperature
@@ -133,7 +144,8 @@ function rewriteKimi(payload) {
 
   const summary = summarizePayload(payload, hasTools, rewriteInfo)
 
-  const consoleMsg = `temperature ${String(incomingTemperature)} -> ${String(rewrittenTemperature)}, top_p ${String(incomingTopP)} -> ${String(forcedTopP)}, thinking ${String(incomingThinkingType)} -> ${String(rewrittenThinkingType)}`
+  const modeTag = hasTools ? '[tools]' : '[chat]'
+  const consoleMsg = `${modeTag} temperature ${String(incomingTemperature)} -> ${String(rewrittenTemperature)}, top_p ${String(incomingTopP)} -> ${String(forcedTopP)}, thinking ${String(incomingThinkingType)} -> ${String(rewrittenThinkingType)}`
 
   // Clean up internal key before forwarding
   delete payload.__incomingThinkingType

package/proxy/qwen-proxy.mjs

@@ -72,7 +72,16 @@ function summarizePayload(payload, hasTools, rewriteInfo) {
 }
 
 function rewriteQwen(payload) {
-  const hasTools = Array.isArray(payload.tools) && payload.tools.length > 0
+  // Determine if a tool is actually being invoked:
+  // - tool_choice is set and not "none"
+  // - OR there is a "tool" role message in the conversation
+  const messages = Array.isArray(payload.messages) ? payload.messages : []
+  const hasToolRole = messages.some((message) => message?.role === 'tool')
+  const toolChoice = payload.tool_choice
+  const hasActiveToolCall =
+    hasToolRole ||
+    (toolChoice !== undefined && toolChoice !== 'none' && toolChoice !== null)
+  const hasTools = hasActiveToolCall
   const incomingEnableThinking = payload.enable_thinking
 
   if (disableThinkingWithTools && hasTools) {
@@ -91,7 +100,8 @@ function rewriteQwen(payload) {
     rewrittenEnableThinking
   })
 
-  const consoleMsg = `tools=${String(hasTools)} enable_thinking=${String(incomingEnableThinking)} -> ${
+  const modeTag = hasTools ? '[tools]' : '[chat]'
+  const consoleMsg = `${modeTag} enable_thinking=${String(incomingEnableThinking)} -> ${
     hasTools && disableThinkingWithTools ? 'false' : '<deleted>'
   }, model=${payload.model ?? '?'}`