KyuRish/trading212-mcp-server

Connect your AI assistant to your Trading 212 brokerage account. Ask questions about your portfolio, place trades, manage pies, and analyze dividends - all through natural language.
Works with **Claude Desktop, Claude Code, ChatGPT, Gemini, Cursor, Windsurf**, and any client that supports the Model Context Protocol (MCP).
Why this server?
- **32 tools** covering the full Trading 212 API, plus 4 analytics tools that combine multiple API calls into actionable insights
- **Smart rate limiting** - reads T212's rate limit headers, auto-waits, and retries on 429 (up to 3 times). No rate limit errors leak to your AI
- **Zero config** - install from PyPI, add your API key, done. No Docker, no database, no Redis
- **Typed responses** - every tool returns structured Pydantic models, not raw JSON
- **Paper trading** - set ENVIRONMENT=demo to test with virtual money first
What can it do?
| Category | Tools | Examples |
|----------|-------|---------|
| **Analytics** | Portfolio summary, performance, dividends, activity | "Show me my portfolio P&L" |
| **Trading** | Market, limit, stop, stop-limit orders | "Buy 5 shares of AAPL" |
| **Portfolio** | Positions, cash balance, account info | "What's my cash balance?" |
| **Pies** | Create, update, duplicate, delete pies | "Show my pie allocations" |
| **Market Data** | Instrument search, exchange schedules | "Search for Tesla" |
| **History** | Past orders, dividends, transactions, CSV exports | "Show my dividend history" |
Analytics tools
These combine multiple API calls into single high-level responses:
- **fetch_portfolio_summary** - Complete snapshot: total value, P&L, cash, top holdings, allocation
- **fetch_portfolio_performance** - Per-position returns with dividends, best/worst performers
- **fetch_dividend_summary** - Income analysis grouped by ticker and month
- **fetch_recent_activity** - Combined timeline of trades and transactions
Quick start
Install
Or via pip:
Get your API credentials
From the Trading 212 app: **Settings > API (Beta)**. You need both the API Key and Secret - the server uses Basic Auth.
Connect to Claude Desktop
Add to claude_desktop_config.json (Windows: %APPDATA%\Claude\, Mac: ~/Library/Application Support/Claude/):
Connect to Claude Code
Then set the environment variables in your shell or .env file.
Other clients (Cursor, Windsurf, ChatGPT, etc.)
Same command and env vars - configure per your client's MCP docs. Set ENVIRONMENT to demo for paper trading.
From source
claude_desktop_config.json for source installs
All 32 tools
Analytics (composite)
| Tool | Description |
|------|-------------|
| fetch_portfolio_summary | Complete portfolio snapshot with P&L and allocations |
| fetch_portfolio_performance | Per-position returns, dividends, best/worst performers |
| fetch_dividend_summary | Dividend income by ticker and month |
| fetch_recent_activity | Combined timeline of trades and transactions |
Account
| Tool | Description |
|------|-------------|
| fetch_account_info | Account metadata (currency, ID) |