greptile-mcp
Greptile
Documentation
🚀 Greptile MCP Server - TypeScript Edition
A modern, TypeScript-powered MCP (Model Context Protocol) server that provides AI-powered code search and querying capabilities through the Greptile API. Built with the official MCP SDK and designed for seamless integration with AI tools like Claude Desktop, Continue, and other MCP-compatible clients.
✨ Features
🔥 Zero-Installation Experience
# Start immediately with npx - no setup required!
npx greptile-mcp-server --api-key=xxx --github-token=yyy🧠 AI-Powered Code Understanding
- Natural Language Queries: Ask questions about codebases in plain English
- Deep Code Analysis: Understand architecture, patterns, and implementation details
- Cross-Repository Insights: Compare patterns and approaches across multiple codebases
- Session Continuity: Build understanding progressively through conversation
⚡ Modern Architecture
- Official MCP SDK: Built with TypeScript MCP SDK for full protocol compliance
- Streaming Support: Real-time responses with Server-Sent Events
- Type Safety: Full TypeScript integration with comprehensive type definitions
- Plugin Architecture: Extensible design for custom tools and integrations
🛠️ Developer Experience
- NPX Ready: Install and run with a single command
- Auto-Configuration: Intelligent configuration detection and validation
- Interactive Setup: Guided setup wizard for first-time users
- Comprehensive Help: Built-in documentation and usage examples
🚀 Quick Start
Prerequisites
- Node.js 18+ (for optimal performance)
- Greptile API Key - Get yours at app.greptile.com
- GitHub Token - Generate at github.com/settings/tokens with `repo` permissions
Instant Start
# Start immediately (will prompt for credentials if not set)
npx greptile-mcp-server
# With inline credentials
npx greptile-mcp-server --api-key=your_key --github-token=your_token
# Interactive setup wizard
npx greptile-mcp-server init
# Test connectivity
npx greptile-mcp-server testEnvironment Setup
Option 1: .env File (Recommended for local development)
Create a `.env` file in your project root:
GREPTILE_API_KEY=your_greptile_api_key_here
GITHUB_TOKEN=your_github_personal_access_token_here
GREPTILE_BASE_URL=https://api.greptile.com/v2 # OptionalOption 2: System Environment Variables
Linux/macOS (Bash/Zsh):
# Current session
export GREPTILE_API_KEY="your_api_key_here"
export GITHUB_TOKEN="your_github_token_here"
# Permanent (add to ~/.bashrc or ~/.zshrc)
echo 'export GREPTILE_API_KEY="your_api_key_here"' >> ~/.bashrc
echo 'export GITHUB_TOKEN="your_github_token_here"' >> ~/.bashrc
source ~/.bashrcWindows PowerShell:
# Current session
$env:GREPTILE_API_KEY="your_api_key_here"
$env:GITHUB_TOKEN="your_github_token_here"
# Permanent
setx GREPTILE_API_KEY "your_api_key_here"
setx GITHUB_TOKEN "your_github_token_here"
# Note: Restart terminal after using setxWindows Command Prompt:
# Current session
set GREPTILE_API_KEY=your_api_key_here
set GITHUB_TOKEN=your_github_token_here
# Permanent
setx GREPTILE_API_KEY "your_api_key_here"
setx GITHUB_TOKEN "your_github_token_here"API Key and Token Setup
Greptile API Key:
1. Visit Greptile Settings
2. Generate a new API key
3. Copy the key to your environment
GitHub Token:
1. Visit GitHub Settings > Personal Access Tokens
2. Create a "Fine-grained personal access token" for better security
3. Grant `repo` permissions for repositories you want to index
4. Copy the token to your environment
🔧 MCP Client Integration
Claude Desktop
Add to your `claude_desktop_config.json`:
{
"mcpServers": {
"greptile": {
"command": "npx",
"args": ["greptile-mcp-server"],
"env": {
"GREPTILE_API_KEY": "your_api_key",
"GITHUB_TOKEN": "your_github_token"
}
}
}
}Continue IDE Extension
Add to your Continue configuration:
{
"contextProviders": [
{
"name": "greptile-mcp",
"type": "mcp",
"serverName": "greptile",
"command": ["npx", "greptile-mcp-server"]
}
]
}Other MCP Clients
The server uses standard MCP protocol and works with any MCP-compatible client:
# Generic MCP client connection
your-mcp-client connect --command "npx greptile-mcp-server"🛠️ Available Tools
1. `greptile_help`
Get comprehensive documentation and usage examples.
{
"name": "greptile_help"
}2. `index_repository`
Index a repository to make it searchable.
{
"name": "index_repository",
"arguments": {
"remote": "github",
"repository": "microsoft/vscode",
"branch": "main",
"reload": true
}
}3. `query_repository`
Query repositories with natural language.
{
"name": "query_repository",
"arguments": {
"query": "How is authentication implemented in this codebase?",
"repositories": [
{
"remote": "github",
"repository": "microsoft/vscode",
"branch": "main"
}
],
"stream": false,
"session_id": "optional-session-id"
}
}4. `get_repository_info`
Get information about indexed repositories.
{
"name": "get_repository_info",
"arguments": {
"remote": "github",
"repository": "microsoft/vscode",
"branch": "main"
}
}📖 Usage Examples
Basic Workflow
# 1. Start the server
npx greptile-mcp-server
# 2. In your MCP client, index a repository
{
"tool": "index_repository",
"arguments": {
"remote": "github",
"repository": "microsoft/vscode",
"branch": "main"
}
}
# 3. Query the codebase
{
"tool": "query_repository",
"arguments": {
"query": "How does VS Code handle file watching?",
"repositories": [{"remote": "github", "repository": "microsoft/vscode", "branch": "main"}]
}
}Advanced Session-Based Exploration
// Start with architecture overview
const session = "exploration-session-1";
// Query 1: High-level understanding
{
"tool": "query_repository",
"arguments": {
"query": "What is the overall architecture of this codebase?",
"session_id": session,
"repositories": [...]
}
}
// Query 2: Deep dive (builds on previous context)
{
"tool": "query_repository",
"arguments": {
"query": "How do the main components we just discussed interact with each other?",
"session_id": session // Same session for continuity
}
}
// Query 3: Implementation details
{
"tool": "query_repository",
"arguments": {
"query": "Show me the specific implementation of the component interaction patterns",
"session_id": session
}
}🔀 Migration from Python Version
The TypeScript version maintains full compatibility with the Python implementation while adding significant improvements:
What's New
- ✅ Official MCP SDK: Standards-compliant implementation
- ✅ NPX Distribution: Zero-installation experience
- ✅ Better Performance: V8 engine advantages for I/O operations
- ✅ Type Safety: Full TypeScript integration
- ✅ Modern Tooling: ESLint, Prettier, comprehensive testing
- ✅ Enhanced CLI: Interactive setup and better UX
Migration Steps
# Old Python usage
python -m src.main
# New TypeScript usage
npx greptile-mcp-server
# Same MCP tools and API compatibility
# No changes needed in MCP client configurations🚨 Troubleshooting
Testing Your Setup
Always test your configuration after setup:
npx greptile-mcp-server testCommon Issues
❌ "Environment variables missing"
Problem: Server can't find your API keys
Solutions:
- Restart your terminal after setting permanent environment variables
- Verify environment variables are set:
# Linux/macOS
echo $GREPTILE_API_KEY
echo $GITHUB_TOKEN
# Windows PowerShell
echo $env:GREPTILE_API_KEY
echo $env:GITHUB_TOKEN- Try using inline credentials:
GREPTILE_API_KEY="your_key" GITHUB_TOKEN="your_token" npx greptile-mcp-server❌ "GitHub token validation failed"
Problem: GitHub token is invalid or has insufficient permissions
Solutions:
- Ensure your token has `repo` permissions
- Generate a new token at GitHub Settings
- For better security, use "Fine-grained personal access tokens"
- Check token hasn't expired
❌ "Greptile API authentication failed"
Problem: Greptile API key is invalid or expired
Solutions:
- Get a new API key from Greptile Settings
- Verify the key is correctly copied (no extra spaces)
- Check if your API key has expired
❌ "Cannot find module" or import errors
Problem: NPX cache issues or incomplete installation
Solutions:
- Clear NPX cache: `npx clear-npx-cache`
- Force fresh install: `npx greptile-mcp-server@latest`
- Check Node.js version (requires Node 18+)
❌ MCP client connection issues
Problem: Claude Desktop or other MCP client can't connect
Solutions:
- Verify MCP server configuration syntax
- Check Claude Desktop logs for detailed error messages
- Ensure environment variables are accessible to the MCP client
- Try running the server manually first to verify it works
Getting Help
- Run `npx greptile-mcp-server init` for interactive setup
- Run `npx greptile-mcp-server test` for detailed diagnostics
- Check the Greptile Documentation for API-specific issues
- Visit MCP Documentation for client integration help
❓ Frequently Asked Questions
Q: Do I need to install anything locally to use this?
A: No! The server runs via NPX with zero installation required. Just run `npx greptile-mcp-server` and it will download and run automatically.
Q: Can I use this with any MCP-compatible client?
A: Yes! This server implements the standard Model Context Protocol and works with Claude Desktop, MCP CLI tools, and any other MCP-compatible client.
Q: How do I index private repositories?
A: Ensure your GitHub token has `repo` permissions for private repositories. The token needs access to read the repositories you want to index.
Q: What's the difference between .env files and environment variables?
A:
- .env files are great for local development - they only work in the directory where the file exists
- Environment variables are system-wide and work everywhere, making them better for global usage with npx
Q: How much does it cost to use Greptile?
A: Greptile pricing depends on your usage. Check Greptile's pricing page for current rates. This MCP server itself is free and open-source.
Q: Can I use this with multiple repositories?
A: Yes! You can index multiple repositories and query across all of them. Use the `index_repository` tool for each repository you want to add.
Q: How long does it take to index a repository?
A: Indexing time varies by repository size. Small repos (< 1000 files) typically take 1-2 minutes, while larger repos may take 10-15 minutes. You can check status with the `get_repository_info` tool.
Q: Is my code data secure?
A: Your code is processed by Greptile's API according to their security and privacy policies. Check Greptile's security documentation for details about data handling and retention.
Q: Can I run this on Windows?
A: Yes! The server works on Windows, macOS, and Linux. Use the platform-specific environment variable setup instructions above.
Q: Why do I get "command not found" errors?
A: This usually means:
- NPX is not installed (install Node.js which includes NPX)
- Your PATH doesn't include Node.js binaries
- There's a typo in the command (it's `npx greptile-mcp-server` not `npx @greptile/mcp-server`)
🏗️ Development
Local Development
# Clone and setup
git clone https://github.com/greptile/mcp-server.git
cd mcp-server
npm install
# Development with hot reload
npm run dev
# Build for production
npm run build
# Run tests
npm test
# Type checking
npm run typecheck
# Linting and formatting
npm run lint
npm run formatProject Structure
src/
├── cli.ts # NPX CLI interface
├── server.ts # Core MCP server implementation
├── index.ts # Module exports
├── clients/
│ └── greptile.ts # Greptile API client
├── types/
│ └── index.ts # TypeScript type definitions
└── utils/
└── index.ts # Utility functions
tests/
├── unit/ # Unit tests
└── integration/ # Integration testsBuild Configuration
- TypeScript: ES2022 target with strict mode
- Build Tool: tsup for dual ESM/CJS output
- Testing: Mocha + Chai with TypeScript support
- Code Quality: ESLint + Prettier with TypeScript rules
🔧 Configuration Options
CLI Arguments
npx greptile-mcp-server \
--api-key="your_key" \
--github-token="your_token" \
--base-url="https://api.greptile.com/v2" \
--repositories='[{"remote":"github","repository":"owner/repo","branch":"main"}]' \
--stream=true \
--timeout=60000 \
--verboseEnvironment Variables
| Variable | Description | Default |
|---|---|---|
| `GREPTILE_API_KEY` | Greptile API key | Required |
| `GITHUB_TOKEN` | GitHub personal access token | Required |
| `GREPTILE_BASE_URL` | API base URL | `https://api.greptile.com/v2` |
Configuration File (Optional)
Create `greptile.config.js`:
export default {
apiKey: process.env.GREPTILE_API_KEY,
githubToken: process.env.GITHUB_TOKEN,
repositories: [
{ remote: 'github', repository: 'owner/repo', branch: 'main' }
],
features: {
streaming: true,
orchestration: true,
flowEnhancement: true
}
};🚀 Deployment
Smithery Cloud Deployment
Deploy instantly to Smithery with zero configuration:
# Install Smithery CLI
npm install -g smithery
# Deploy from repository
smithery deploy
# Or deploy with custom configuration
smithery deploy --config smithery.yaml
# Monitor deployment
smithery status
smithery logsDocker Deployment
# Build Docker image
docker build -t greptile-mcp .
# Run with environment variables
docker run -e GREPTILE_API_KEY=your_key \
-e GITHUB_TOKEN=your_token \
-p 8080:8080 \
greptile-mcp
# Or build Smithery-optimized image
npm run smithery:buildEnvironment Variables for Deployment
# Required
GREPTILE_API_KEY=your_greptile_api_key
GITHUB_TOKEN=your_github_token
# Optional
GREPTILE_BASE_URL=https://api.greptile.com/v2
TRANSPORT=stdio
HOST=0.0.0.0
PORT=8080Cloud Platforms
- Smithery: One-click deployment with `smithery deploy`
- Railway: Connect GitHub repo, set environment variables
- Render: Use `npm start` as start command
- Heroku: Standard Node.js deployment
- DigitalOcean App Platform: Docker or buildpack deployment
🚦 Performance & Benchmarks
Startup Performance
- Cold Start: < 2 seconds
- Memory Usage: ~50MB base footprint
- Concurrent Requests: Handles 100+ concurrent MCP tool calls
API Performance
- Query Response: Typically 1-3 seconds
- Streaming: Real-time chunk delivery
- Repository Indexing: Varies by repository size (usually 30s-5min)
Compared to Python Version
- 40% Faster Startup - V8 vs Python runtime
- 60% Smaller Container - Node.js vs Python base images
- 30% Better Memory Efficiency - V8 garbage collection
- Native Streaming - Better SSE performance
🛡️ Security & Best Practices
Token Security
- Environment Variables: Store tokens in environment, not code
- Minimal Permissions: Use GitHub tokens with only required `repo` permissions
- Token Rotation: Regularly rotate API keys and tokens
- Local Storage: Never commit tokens to version control
Network Security
- HTTPS Only: All API communications use HTTPS
- Request Validation: Input validation and sanitization
- Rate Limiting: Built-in retry logic with exponential backoff
- Error Handling: Comprehensive error handling without token exposure
🤝 Contributing
We welcome contributions! Please see our Contributing Guide for details.
Development Workflow
1. Fork the repository
2. Create a feature branch (`git checkout -b feature/amazing-feature`)
3. Make your changes with tests
4. Run the test suite (`npm test`)
5. Ensure code quality (`npm run lint`)
6. Commit your changes (`git commit -m 'Add amazing feature'`)
7. Push to the branch (`git push origin feature/amazing-feature`)
8. Open a Pull Request
📄 License
This project is licensed under the MIT License - see the LICENSE file for details.
🙏 Acknowledgments
- Anthropic for the Model Context Protocol specification
- Greptile for the powerful code analysis API
- TypeScript Community for excellent tooling and ecosystem
- MCP Community for protocol development and feedback
📞 Support
- Documentation: docs.greptile.com
- Issues: GitHub Issues
- Discussions: GitHub Discussions
- Discord: MCP Community Discord
Built with ❤️ by the Greptile team • Powered by TypeScript and the Model Context Protocol
Frequently asked questions
What is greptile-mcp?
greptile-mcp is Greptile
How do I install greptile-mcp?
Open the GitHub repository and follow its README. Most MCP servers are added to your client's MCP config, then called by your agent.
Is greptile-mcp open source?
Yes — it is hosted on GitHub at https://github.com/sosacrazy126/greptile-mcp and has 8 stars.
Related MCP tools
🔥 Official Firecrawl MCP Server - Adds powerful web scraping and search to Cursor, Claude and any other LLM clients.
Use any LLMs (Large Language Models) for Deep Research. Support SSE API and MCP server.
Enhanced MCP server for interactive user feedback and command execution in AI-assisted development, featuring dual interface support (Web UI and Desktop Application) with intelligent environment detection and cross-platform compatibility.
A powerful Zotero AI and MCP plugin with ChatGPT, Gemini 3.7, Claude Fable 5, Claude Opus 5, DeepSeek V4, Grok, OpenRouter, Kimi k3, GLM 5.3, SiliconFlow, GPT-oss, Gemma 4, Qwen 3.8
Connect your browser to AI models. Just use Dia on Chrome, Arc or Firefox.
文颜 MCP Server 可以让 AI 自动将 Markdown 文章排版后发布至微信公众号。
Run your own MCP server? See who uses it and what to fix.
Measure it with TrackMCP