copilot-custom-endpoint 1.3.10 → 1.3.12

package/README.md

@@ -107,6 +107,8 @@ For the full pricing comparison (cached rates, full Copilot roster, footnotes, s
 
 ## Companion tools
 
+> **ℹ️ These are third-party tools — not built into the custom endpoints or proxies in this repo.** Each one must be installed, configured, and (where applicable) billed for **separately**, directly with its own provider. Nothing here is bundled, proxied, or auto-configured by `copilot-custom-endpoint` or the per-model setups above. The entries below are just pointers to tools the author has found useful alongside the model configs.
+
 These work alongside the providers above and fill gaps that VS Code's built-in tool surface doesn't cover natively.
 
 ### 🎬 [Video Context MCP](https://www.videocontextmcp.com/) — _video understanding for AI coding assistants_
@@ -120,6 +122,19 @@ VS Code's built-in `view_image` tool only accepts **static images** (PNG, JPG, G
 - **Answers natural-language questions** about the video grounded in actual frames: "what does the speaker click in the last 30 seconds?", "summarize the demo", "find the frame where the error appears".
 - **Extras:** timestamp search, audio transcription with speaker diarization, and video metadata (resolution, duration, codec).
 
+### 🪣 [Bitbucket MCP](https://bitbucketmcp.tugudush.com/) — _secure, read-only Bitbucket access for VS Code Copilot, Cursor, and Claude Code_
+
+GitHub ships a first-party MCP server (and it's even bundled into Copilot), so asking "what's open in my org's repos / show me PR #123" works seamlessly on github.com. **Bitbucket has no equivalent** — Atlassian hasn't shipped one — which leaves Bitbucket Cloud users copy-pasting PR URLs, diffs, and file contents into chat by hand.
+
+**Bitbucket MCP** is a small MCP server that closes that gap. It works with **VS Code GitHub Copilot, Cursor, and Claude Code** out of the box, and:
+
+- **38 tools across 8 categories** — repositories, pull requests, branches & commits, diffs & comparisons, CI/CD pipelines, issues, code search, and users.
+- **Read-only by design** — `makeRequest()` blocks all non-GET requests at runtime, so no write, delete, or modify operation is possible.
+- **Token-friendly output** — every tool supports `text`, `json`, and `toon` (Token-Oriented Object Notation) formats; `toon` cuts LLM token consumption by 30–60% on large PR/commit lists.
+- **JMESPath filtering** on all 38 tools, so you can trim responses (e.g. only open PRs, or just title + author) before they hit the model.
+- **One-call PR context** — `bb_get_context` bundles PR metadata, diffstat, CI statuses, and comments in a single request.
+- **Drop-in install** — `npm install -g @tugudush/bitbucket-mcp` plus a short `.vscode/mcp.json` entry, authenticated with a Bitbucket API token + your Atlassian email.
+
 ## Need help?
 
 - **Per-model issues:** check the troubleshooting section at the bottom of each model's doc.

package/docs/example-config.md

@@ -14,7 +14,7 @@ Here's a complete, real-world `chatLanguageModels.json` that combines **all the
     "models": [
       {
         "id": "qwen3.7-max",
-        "name": "Qwen 3.7 Max",
+        "name": "Qwen 3.7 Max (text)",
         "url": "https://dashscope-intl.aliyuncs.com/compatible-mode/v1/chat/completions",
         "toolCalling": true,
         "vision": false,
@@ -25,7 +25,7 @@ Here's a complete, real-world `chatLanguageModels.json` that combines **all the
       },
       {
         "id": "qwen3.7-plus",
-        "name": "Qwen 3.7 Plus",
+        "name": "Qwen 3.7 Plus (vision)",
         "url": "https://dashscope-intl.aliyuncs.com/compatible-mode/v1/chat/completions",
         "toolCalling": true,
         "vision": true,
@@ -44,7 +44,7 @@ Here's a complete, real-world `chatLanguageModels.json` that combines **all the
     "models": [
       {
         "id": "kimi-k2.6",
-        "name": "Kimi K2.6",
+        "name": "Kimi K2.6 (vision)",
         "url": "http://127.0.0.1:3457/v1/chat/completions",
         "requestBody": {
           "temperature": 1
@@ -65,7 +65,7 @@ Here's a complete, real-world `chatLanguageModels.json` that combines **all the
     "models": [
       {
         "id": "mimo-v2.5-pro",
-        "name": "MiMo V2.5 Pro",
+        "name": "MiMo V2.5 Pro (text)",
         "url": "https://api.xiaomimimo.com/v1/chat/completions",
         "toolCalling": true,
         "vision": false,
@@ -80,7 +80,7 @@ Here's a complete, real-world `chatLanguageModels.json` that combines **all the
       },
       {
         "id": "mimo-v2.5",
-        "name": "MiMo V2.5",
+        "name": "MiMo V2.5 (vision)",
         "url": "https://api.xiaomimimo.com/v1/chat/completions",
         "toolCalling": true,
         "vision": true,
@@ -95,7 +95,7 @@ Here's a complete, real-world `chatLanguageModels.json` that combines **all the
       },
       {
         "id": "mimo-v2-flash",
-        "name": "MiMo V2 Flash",
+        "name": "MiMo V2 Flash (text)",
         "url": "https://api.xiaomimimo.com/v1/chat/completions",
         "toolCalling": true,
         "vision": false,
@@ -118,7 +118,7 @@ Here's a complete, real-world `chatLanguageModels.json` that combines **all the
     "models": [
       {
         "id": "MiniMax-M3",
-        "name": "MiniMax M3",
+        "name": "MiniMax M3 (vision)",
         "url": "https://api.minimax.io/v1/chat/completions",
         "toolCalling": true,
         "vision": true,
@@ -142,7 +142,7 @@ Here's a complete, real-world `chatLanguageModels.json` that combines **all the
     "models": [
       {
         "id": "glm-5.1",
-        "name": "GLM 5.1 (flagship)",
+        "name": "GLM 5.1 (text)",
         "url": "https://api.z.ai/api/paas/v4/chat/completions",
         "toolCalling": true,
         "vision": false,
@@ -158,7 +158,7 @@ Here's a complete, real-world `chatLanguageModels.json` that combines **all the
 
       {
         "id": "glm-5v-turbo",
-        "name": "GLM 5V Turbo (vision flagship)",
+        "name": "GLM 5V Turbo (vision)",
         "url": "https://api.z.ai/api/paas/v4/chat/completions",
         "toolCalling": true,
         "vision": true,

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`                               |
 
@@ -165,12 +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": "1.3.10",
+  "version": "1.3.12",
   "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 ?? '?'}`