Portero
A self-hosted MCP gateway that aggregates multiple servers into a single endpoint with features for data anonymization and granular permission policies. It integrates a Telegram bot for real-time 2FA approvals of sensitive operations and secure remote access.
README
Portero
A self-hosted MCP (Model Context Protocol) gateway that sits between Claude Code and multiple MCP servers, providing:
- MCP Aggregation — Connect multiple MCPs and expose them as one unified endpoint
- Data Anonymization — Bidirectional fake↔real data replacement for privacy
- 2FA Approvals — Telegram bot for approving sensitive operations
- Permission Policies — Allow/deny/require-approval per tool
- Remote Access — HTTPS endpoint accessible from anywhere
Architecture
┌─────────────────────────────────────────────────────────────┐
│ TELEGRAM BOT │
│ /status, /grant, /revoke, approval callbacks │
└─────────────────────┬───────────────────────────────────────┘
│
┌─────────────────────▼───────────────────────────────────────┐
│ PORTERO │
│ ┌────────────────────────────────────────────────────────┐│
│ │ HTTP Server (Express) ││
│ │ - POST /mcp/message (JSON-RPC, Bearer auth) ││
│ │ - GET /health ││
│ └────────────────────────────────────────────────────────┘│
│ ┌────────────────────────────────────────────────────────┐│
│ │ Middleware Pipeline ││
│ │ 1. Anonymization (fake→real on requests) ││
│ │ 2. Policy Check (allow/deny/require-approval) ││
│ │ 3. 2FA Gate (wait for Telegram approval if needed) ││
│ │ 4. Route to child MCP ││
│ │ 5. Anonymization (real→fake on responses) ││
│ └────────────────────────────────────────────────────────┘│
└─────────────────────────────────────────────────────────────┘
│ stdio
┌─────────────┼─────────────┐
▼ ▼ ▼
┌─────────┐ ┌─────────┐ ┌─────────┐
│ MCP 1 │ │ MCP 2 │ │ MCP 3 │
│(github) │ │(filesys)│ │(google) │
└─────────┘ └─────────┘ └─────────┘
Prerequisites
- Node.js 20+ (LTS recommended)
- Telegram Bot (create via @BotFather)
- Your Telegram Chat ID (get from @userinfobot)
Quick Start
1. Clone and Install
git clone <your-repo-url>
cd portero
npm install
2. Configure Environment
cp .env.example .env
# Edit .env with your settings
Required settings in .env:
# Generate a secure token
BEARER_TOKEN=$(openssl rand -hex 32)
# Get from @BotFather
TELEGRAM_BOT_TOKEN=123456:ABC-DEF1234ghIkl-zyx57W2v1u123ew11
# Get from @userinfobot
TELEGRAM_ADMIN_CHAT_ID=123456789
# Your real info (for anonymization)
REAL_NAME="Your Name"
REAL_EMAIL="your@email.com"
3. Configure MCP Servers
Edit config/mcps.json to define which MCP servers to connect:
{
"mcps": [
{
"name": "filesystem",
"command": "npx",
"args": ["-y", "@modelcontextprotocol/server-filesystem", "/workspace"],
"env": {}
},
{
"name": "github",
"command": "npx",
"args": ["-y", "@modelcontextprotocol/server-github"],
"env": {
"GITHUB_PERSONAL_ACCESS_TOKEN": "${GITHUB_TOKEN}"
}
}
]
}
4. Configure Google Workspace (Optional)
To add Gmail, Calendar, and Drive integration via workspace-mcp:
- Create a Google Cloud project at console.cloud.google.com
- Enable APIs: Gmail API, Google Calendar API, Google Drive API
- Create OAuth 2.0 credentials: APIs & Services → Credentials → Create Credentials → OAuth client ID → Desktop app
- Set environment variables in
.env:GOOGLE_OAUTH_CLIENT_ID=your-client-id.apps.googleusercontent.com GOOGLE_OAUTH_CLIENT_SECRET=your-client-secret - First run: The workspace-mcp server will open a browser for OAuth consent. Approve the requested scopes.
- Headless / Docker: Run once locally to complete the OAuth flow, then copy the token cache into the container.
If GOOGLE_OAUTH_CLIENT_ID and GOOGLE_OAUTH_CLIENT_SECRET are not set, Portero will skip the Google MCP and start without it.
The Google tools appear as google/send_email, google/list_events, google/search_files, etc. Write operations (send, create, delete) require Telegram approval; reads are allowed by default. See config/policies.json for the full list.
5. Configure Data Anonymization
Edit config/replacements.json to define fake↔real mappings:
{
"replacements": [
{
"fake": "John Doe",
"real": "${REAL_NAME}",
"bidirectional": true
},
{
"fake": "john@example.com",
"real": "${REAL_EMAIL}",
"bidirectional": true,
"caseSensitive": false
}
]
}
6. Configure Policies
Edit config/policies.json to set permission rules:
{
"policies": {
"github/create_issue": "allow",
"github/create_pull_request": "require-approval",
"filesystem/write_file": "require-approval",
"filesystem/read_file": "allow",
"filesystem/delete_file": "deny",
"*": "allow"
},
"defaultPolicy": "allow"
}
7. Generate SSL Certificates (Optional)
./scripts/generate-certs.sh
Or skip SSL for local testing (uses HTTP).
8. Start the Gateway
# Development mode (with hot reload)
npm run dev
# Production mode
npm run build
npm start
Docker Deployment
# Build and start with docker-compose
docker-compose up -d
# View logs
docker-compose logs -f
# Stop
docker-compose down
Connect from Claude Code
Add to your Claude Code MCP configuration:
{
"mcpServers": {
"portero": {
"transport": "http",
"url": "https://your-server:8443/mcp/message",
"headers": {
"Authorization": "Bearer your-bearer-token-here"
}
}
}
}
Add to Claude Code system prompt:
Your identity:
- Name: John Doe
- Email: john@example.com
Use these when asked for personal information.
Telegram Bot Commands
Once running, message your bot:
/status- Show connected MCPs, active grants, pending approvals/grant <pattern> <duration>- Grant temporary access- Examples:
/grant github/* 30m,/grant * 1h
- Examples:
/revoke- Revoke all active grants/pending- Show pending approval requests/logs- Show recent audit logs/help- Show all commands
How 2FA Approval Works
- Claude Code calls a tool (e.g.,
github/create_pull_request) - Gateway intercepts the request
- Policy engine checks: requires approval
- Telegram bot sends you a message with Approve/Deny buttons
- You click a button
- Gateway proceeds or blocks based on your decision
- Response returns to Claude Code
Configuration Reference
Data Anonymization
Replacements support:
- Bidirectional — Replace in both directions (fake↔real)
- One-way — Replace only fake→real, use
responseReplacementfor responses - Case sensitivity — Set
caseSensitive: falsefor case-insensitive matching
Permission Policies
Policy actions:
allow— Allow without approvaldeny— Block completelyrequire-approval— Request Telegram approval
Patterns support wildcards:
github/*— All GitHub tools*/delete_*— All delete operations*— All tools
Temporary Grants
Skip approval for a limited time:
/grant github/* 30m # Grant GitHub access for 30 minutes
/grant * 1h # Grant all access for 1 hour
/revoke # Revoke all grants immediately
Security Considerations
-
Bearer Token — Generate a strong random token:
openssl rand -hex 32 -
SSL/TLS — Use HTTPS in production (Let's Encrypt, self-signed, or reverse proxy)
-
Telegram — Only your admin chat ID can control the bot
-
Firewall — Restrict gateway port (8443) to authorized IPs
-
Environment Variables — Never commit
.envto git
Development
Project Structure
portero/
├── src/
│ ├── index.ts # Entry point
│ ├── config/ # Config loader
│ ├── gateway/ # HTTP server & handler
│ ├── mcp/ # MCP client management
│ ├── middleware/ # Anonymizer, policy, approval
│ ├── telegram/ # Telegram bot
│ ├── db/ # SQLite database
│ └── utils/ # Logger, crypto
├── config/ # JSON config files
├── docker/ # Docker setup
└── scripts/ # Helper scripts
Build Commands
npm run dev # Development with hot reload
npm run build # Compile TypeScript
npm start # Start production build
Database
SQLite database at ./data/gateway.db stores:
- Pending approvals
- Temporary grants
- Audit logs
Troubleshooting
Gateway won't start
- Check Node.js version:
node -v(should be 20+) - Verify
.envfile exists and has all required variables - Check logs in
./logs/combined.log
MCP connection fails
- Verify MCP command is correct in
config/mcps.json - Check MCP is installed:
npx -y @modelcontextprotocol/server-github --version - Check environment variables are set (e.g.,
GITHUB_TOKEN)
Telegram bot not responding
- Verify bot token is correct
- Check admin chat ID matches your Telegram ID
- Ensure bot was started with
/start
Claude Code can't connect
- Verify bearer token matches in Claude Code config
- Check SSL certificates if using HTTPS
- Test with
curl:curl -X POST https://localhost:8443/health
Contributing
Contributions welcome! Please:
- Fork the repository
- Create a feature branch
- Make your changes
- Submit a pull request
License
MIT License - see LICENSE file for details
Support
- GitHub Issues: [Report bugs or request features]
- Documentation: [See PROMPT.md for architecture details]
Built for Claude Code users who want privacy, security, and control over their MCP connections.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.