VS Code extension that spins up a local Model Context Protocol (MCP) HTTP server, exposing VS Code’s file, edit, symbol, diagnostics and shell features so AI coding agents (e.g. Claude) can work directly against an open workspace.
https://github.com/juehang/vscode-mcp-serverStop context-switching between your AI coding assistant and VS Code. This MCP server turns your editor into a programmable workspace that AI agents can navigate, edit, and understand with the same precision you do.
You're already using AI for code generation, but every time your assistant needs to understand your project structure or make targeted edits, you're copy-pasting files and manually implementing changes. That workflow breaks down fast on real codebases.
VS Code MCP Server solves this by exposing your editor's core capabilities—file operations, symbol search, diagnostics, and shell access—through the Model Context Protocol. Your AI agent gets direct access to the same workspace navigation and editing tools you use daily.
Intelligent Code Navigation
Precise Code Editing
Context-Efficient Exploration Instead of dumping entire files into context, your AI can:
Extend Existing AI Coding Agents: Perfect for Claude Code or similar tools. Enable only the symbol and diagnostics tools while letting your existing agent handle basic file operations—no capability duplication, just enhanced VS Code integration.
Automated Code Reviews: Your AI can get_diagnostics_code
across the workspace, identify issues, and propose fixes with line-precise edits through replace_lines_code
.
Intelligent Refactoring: Search for symbol usage with search_symbols_code
, understand definitions with get_symbol_definition_code
, then make coordinated changes across multiple files.
Codebase Onboarding: New team members (or AIs) can rapidly understand project structure through get_document_symbols_code
and symbol navigation without drowning in irrelevant code.
Install from the VS Code Marketplace, configure your Claude Desktop with a simple JSON snippet, and you're running. The server operates locally on a configurable port with selective tool enabling—use only what adds value to your existing workflow.
{
"mcpServers": {
"vscode-mcp-server": {
"command": "npx",
"args": ["mcp-remote@next", "http://localhost:3000/mcp"]
}
}
}
Security-Conscious: Shell command execution is opt-in, the server only runs locally, and you control which tool categories are enabled through VS Code settings.
Workflow-Friendly: Toggle the server on/off via the status bar. Configure which tools are exposed to avoid conflicts with your existing AI setup.
Your AI agent should understand your codebase structure, navigate it intelligently, and make precise edits that integrate properly with your development environment. VS Code MCP Server makes that possible today.
The 121 GitHub stars and active development show other developers are already using this to supercharge their AI coding workflows. The question isn't whether you need better AI-IDE integration—it's whether you want to keep manually bridging that gap while your productivity suffers.
Get Started: Install from the VS Code Marketplace, add the Claude configuration, and watch your AI agent navigate your codebase like it actually understands your development environment.