mcp -vscode
Give Claude Code direct access to VS Code's code navigation, error checking, and debugging tools through the Model Context Protocol.
claude mcp add --transport stdio malvex-mcp-server-vscode npx github:malvex/mcp-server-vscode
How to use
This MCP server enables AI assistants (via Claude or similar clients) to interact with your VS Code environment through a lightweight bridge. The VS Code extension provides language intelligence features (hover, definitions, references, symbol searches, diagnostics) and debugging controls, while the MCP Standalone Server translates between the editor’s stdio workflow and an HTTP API that clients can query. When running, the extension exposes an HTTP API on port 8991; the MCP Standalone Server sits in the middle to translate requests between stdio and HTTP so that the AI assistant can issue MCP commands and receive responses.
To use it, install the VS Code extension and start the MCP server (via Claude or directly from your terminal). Claude Desktop (or Claude Code) will connect to the MCP server through the HTTP/stdio bridge, enabling commands like hover, go-to-definition, references, and debugging actions such as starting sessions, setting breakpoints, stepping through code, and inspecting variables. The workflow is designed so Claude can issue high-level language operations and debugging tasks without needing to know VS Code internals.
Available tools are organized into three categories: Language Intelligence (e.g., hover, definition, references, symbolSearch, workspaceSymbols, diagnostics, callHierarchy), Debug Tools (breakpoint management, session control, runtime control, and inspection/evaluation utilities), and a Refactoring tool (rename symbol across the workspace). Each tool accepts structured parameters and returns results compatible with the MCP protocol, enabling seamless conversational or scripted interactions with your codebase and debugging sessions.
How to install
Prerequisites:
- Visual Studio Code installed
- Node.js and npm/yarn available (for npx usage)
- Claude Desktop (or Claude Code) installed and configured
Step-by-step installation:
# 1) Install the VS Code extension (Alpha)
# You would typically download the .vsix from Releases and install in VS Code.
# In this guide, we rely on the MCP server running via npx as described below.
# 2) Run the MCP server via npx (from a terminal)
npx github:malvex/mcp-server-vscode
# 3) Configure Claude Desktop to connect to the MCP server
# macOS
# ~/Library/Application Support/Claude/claude_desktop_config.json
# Windows
# %APPDATA%\Claude\claude_desktop_config.json
# Example Claude Desktop config entry
{
"mcpServers": {
"vscode": {
"command": "npx",
"args": ["github:malvex/mcp-server-vscode"]
}
}
}
# 4) (Optional) Claude Code CLI integration
claude mcp add-json vscode '{"type":"stdio","command":"npx","args":["github:malvex/mcp-server-vscode"]}' -s user
Notes:
- If you prefer, you can run the MCP server directly from the terminal and connect Claude Desktop via stdio or HTTP as described in the extension’s docs.
- The server uses port 8991 for the HTTP interface exposed by the VS Code extension.
Additional notes
- The MCP server for VS Code bridges stdio and HTTP; ensure firewall/localhost access is allowed to port 8991 when using Claude Desktop.
- The VS Code extension provides various tools; you can access 25+ tools across Language Intelligence, Debugging, and Refactoring categories as documented in the README.
- If Claude cannot connect, verify that VS Code is running with the extension active and that the MCP server reports a running port (e.g., VS Code MCP: 8991).
- For CLI users, you can manually start the MCP server using the npx command shown above and test with a simple HTTP request to http://localhost:8991.
Related MCP Servers
systemprompt-code-orchestrator
MCP server for orchestrating AI coding agents (Claude Code CLI & Gemini CLI). Features task management, process execution, Git integration, and dynamic resource discovery. Full TypeScript implementation with Docker support and Cloudflare Tunnel integration.
mcp-memory-keeper
MCP server for persistent context management in AI coding assistants
mcp-rest-api
A TypeScript-based MCP server that enables testing of REST APIs through Cline. This tool allows you to test and interact with any REST API endpoints directly from your development environment.
talkito
TalkiTo lets developers interact with AI systems through speech across multiple channels (terminal, API, phone). It can be used as both a command-line tool and a Python library.
cco
Real-time audit and approval system for Claude Code tool calls.
mcp-install-instructions-generator
Generate MCP Server Installation Instructions for Cursor, Visual Studio Code, Claude Code, Claude Desktop, Windsurf, ChatGPT, Gemini CLI and more