Bitpanda MCP Server

The official Model Context Protocol (MCP) server for Bitpanda. Connects AI agents — Claude, Cursor, VS Code Copilot, and any MCP-compatible client — directly to the Bitpanda API, giving them secure, read-only access to your asset portfolio, wallet balances, trade history, and real-time market prices.

Built with FastMCP 3.x, Python 3.11+, and uv.

Install the Bitpanda MCP Server

Claude Code

claude mcp add bitpanda -e BITPANDA_API_KEY=your-key -- uvx --from git+https://github.com/bitpanda-labs/bitpanda-mcp bitpanda-mcp

Claude Desktop

Add to ~/Library/Application Support/Claude/claude_desktop_config.json:

{
  "mcpServers": {
    "bitpanda": {
      "command": "uvx",
      "args": ["--from", "git+https://github.com/bitpanda-labs/bitpanda-mcp", "bitpanda-mcp"],
      "env": {
        "BITPANDA_API_KEY": "your-key"
      }
    }
  }
}

Cursor / VS Code / Windsurf / any MCP client

uvx --from git+https://github.com/bitpanda-labs/bitpanda-mcp bitpanda-mcp

Set BITPANDA_API_KEY in your environment or .env file.

Get your Bitpanda API key at web.bitpanda.com/apikey.

What Can You Do With Bitpanda MCP?

• Analyze your portfolio with Claude — ask "what's my BTC allocation?" or "show me my top performing assets this month"

• Get real-time asset prices — BTC, ETH, and hundreds of other assets via natural language

• Review trade and transaction history — "summarize my trades from last quarter" or "show me recent transactions"

• Build AI agents that interact with Bitpanda data — integrate Bitpanda portfolio data into any MCP-compatible AI workflow

• Automate reporting — feed wallet balances and trade history into AI-powered analysis pipelines

Available MCP Tools

All tools are read-only and annotated with readOnlyHint=true. The server never writes to or modifies your Bitpanda account.

MCP Prompts

Self-Hosted Bitpanda MCP Server (HTTP / Docker)

Run the server as a remote HTTP service — multiple users each authenticate per-request with their own Bitpanda API key in the X-Api-Key header. Stateless design supports horizontal scaling.

Start the HTTP server

FASTMCP_TRANSPORT=streamable-http \
FASTMCP_HOST=0.0.0.0 \
FASTMCP_PORT=8000 \
uv run bitpanda-mcp

Or with Docker (multi-stage build — runtime image contains no build tools or source code):

docker build -f ci/docker/Dockerfile -t bitpanda-mcp .
docker run -p 8000:8000 bitpanda-mcp

Connect from any MCP client

{
  "mcpServers": {
    "bitpanda": {
      "url": "http://localhost:8000/mcp",
      "headers": {
        "X-Api-Key": "YOUR_BITPANDA_API_KEY"
      }
    }
  }
}

Health check: GET /healthz → {"status": "ok"}.

Configuration & Environment Variables

Frequently Asked Questions

Do I need a Bitpanda account?
Yes — you need an active Bitpanda account and a Bitpanda API key. Generate one at web.bitpanda.com/apikey.

Is this the official Bitpanda MCP server?
Yes. This repository is maintained by Bitpanda Labs and is the official MCP integration for the Bitpanda platform.

What is the Model Context Protocol (MCP)?
MCP is an open standard that lets AI assistants connect to external data sources and tools in a structured, secure way. It's supported by Claude, Cursor, VS Code, Windsurf, and a growing ecosystem of AI development tools.

Can Claude write to my Bitpanda account or place trades?
No. All tools exposed by this server are strictly read-only. The server cannot place orders, move funds, or modify your account in any way.

Does this work with Claude.ai or ChatGPT?
It works with Claude Code (CLI) and Claude Desktop. ChatGPT and other OpenAI-based tools can connect via the remote HTTP mode if they support MCP or streamable-http. Any MCP-compatible client is supported.

Development

Requires Python 3.11+.

git clone https://github.com/bitpanda-labs/bitpanda-mcp && cd bitpanda-mcp
uv sync
cp .env.example .env  # add your Bitpanda API key

uv run pytest                  # tests (100% coverage enforced)
uv run ruff check src/ tests/  # lint
uv run ruff format src/ tests/ # format
uv build                       # build wheel + sdist

CI runs lint, tests on Python 3.11–3.14, and verifies the wheel installs cleanly.

License

Apache 2.0