Copilot Chat MCP support addition

.gitignore

@@ -24,6 +24,9 @@ env/
 
 # IDE
 .vscode/
+!.vscode/
+.vscode/*
+!.vscode/mcp.json
 .idea/
 *.swp
 *.swo
@@ -84,4 +87,4 @@ site/
 *.log
 *.tmp
 temp/
-tmp/
+tmp/

.vscode/mcp.json

@@ -0,0 +1,12 @@
+{
+  "mcpServers": {
+    "shortkit-ml": {
+      "type": "stdio",
+      "command": ".venv/bin/python",
+      "args": [
+        "-m",
+        "shortcut_detect.mcp_server"
+      ]
+    }
+  }
+}

README.md

@@ -177,13 +177,13 @@ Custom conditions can be registered via `@register_condition("name")`. See [Cond
 
 ## MCP Server
 
-ShortKit-ML ships an [MCP](https://modelcontextprotocol.io/) server so AI assistants (Claude, Cursor, etc.) can call detection tools directly from chat — no Python script required.
+ShortKit-ML ships an [MCP](https://modelcontextprotocol.io/) server so AI assistants (Claude, Cursor, GitHub Copilot in VS Code, etc.) can call detection tools directly from chat — no Python script required.
 
 ```bash
 uv pip install -e ".[mcp]"
 ```
 
-Full setup instructions for **Claude Code** and **Claude Desktop**, available tools, file-based input, and troubleshooting: [**criticaldata.github.io/ShortKit-ML/mcp-server**](https://criticaldata.github.io/ShortKit-ML/mcp-server/).
+Full setup instructions for **Claude Code**, **Claude Desktop**, and **GitHub Copilot in VS Code**, available tools, file-based input, and troubleshooting: [**criticaldata.github.io/ShortKit-ML/mcp-server**](https://criticaldata.github.io/ShortKit-ML/mcp-server/).
 
 ## Paper Benchmark Datasets
 

docs/mcp-server.md

@@ -1,6 +1,6 @@
 # MCP Server
 
-ShortKit-ML ships a [Model Context Protocol (MCP)](https://modelcontextprotocol.io/) server that exposes shortcut detection as callable tools. AI assistants like Claude can run full analyses — generate data, run detectors, compare methods — directly from chat, with no Python script required.
+ShortKit-ML ships a [Model Context Protocol (MCP)](https://modelcontextprotocol.io/) server that exposes shortcut detection as callable tools. AI assistants like Claude and GitHub Copilot in VS Code can run full analyses — generate data, run detectors, compare methods — directly from chat, with no Python script required.
 
 ## Prerequisites
 
@@ -18,8 +18,6 @@ uv run python -m shortcut_detect.mcp_server
 
 The server should start without errors. Press `Ctrl+C` to stop it.
 
----
-
 ## Option 1 — Claude Code (recommended)
 
 Claude Code reads a `.mcp.json` file from the project root. This file is already committed to the repo, so no manual setup is needed after installing the package.
@@ -113,6 +111,36 @@ Once connected, you can run all 19 detection methods on synthetic data directly
 
 ---
 
+## Option 3 — GitHub Copilot in VS Code
+
+GitHub Copilot Chat in VS Code can consume MCP servers directly in Agent mode. This repository includes a checked-in `.vscode/mcp.json` template so the ShortKit-ML server is ready to use once your `.venv` is installed.
+
+### 1. Install dependencies
+
+```bash
+uv pip install -e ".[mcp]"
+```
+
+### 2. Open the workspace in VS Code
+
+Make sure the repo root is open so VS Code can read `.vscode/mcp.json`.
+
+### 3. Start Copilot Chat in Agent mode
+
+Open Copilot Chat, switch the mode dropdown to **Agent**, and click the tools icon. Under the `shortkit-ml` MCP server, you should see the ShortKit-ML tools.
+
+### 4. Try a prompt
+
+```text
+Generate 200 synthetic samples with a linear shortcut and run probe, statistical, and hbac on them.
+```
+
+### 5. If you need to customize the Python path
+
+Edit `.vscode/mcp.json` and point `command` at your workspace `.venv/bin/python` path. Use the same `shortcut_detect.mcp_server` module entry point as Claude Code and Claude Desktop.
+
+---
+
 ## Quick start — how to use the tools
 
 Once connected, the tools are **automatic** — Claude calls them for you based on what you ask. You never invoke them by name. Just describe what you want in plain English.

tests/test_copilot_mcp_config.py

@@ -0,0 +1,21 @@
+"""Tests for the checked-in Copilot / MCP workspace configuration."""
+
+import json
+from pathlib import Path
+
+
+def test_vscode_mcp_config_matches_project_entrypoint():
+    root = Path(__file__).resolve().parents[1]
+    claude_config = json.loads((root / ".mcp.json").read_text(encoding="utf-8"))
+    vscode_config = json.loads((root / ".vscode" / "mcp.json").read_text(encoding="utf-8"))
+
+    claude_server = claude_config["mcpServers"]["shortkit-ml"]
+    vscode_server = vscode_config["mcpServers"]["shortkit-ml"]
+
+    assert claude_server["type"] == "stdio"
+    assert vscode_server["type"] == "stdio"
+    assert claude_server["command"] == vscode_server["command"] == ".venv/bin/python"
+    assert claude_server["args"] == vscode_server["args"] == [
+        "-m",
+        "shortcut_detect.mcp_server",
+    ]