How-to: Connect the BioCypher MCP to your IDE
Note
This page is a starting point and will be expanded. For the full picture on LLM integration, see the LLM Integration Guide.
BioCypher provides a dedicated Model Context Protocol (MCP) server that AI coding assistants (GitHub Copilot, Claude, Cursor, etc.) can use to help you build and debug BioCypher pipelines directly from your IDE.
1. Install an IDE with MCP support
Any IDE that supports MCP works, for example VSCode with the GitHub Copilot extension.
2. Add the BioCypher MCP server
Add the following to your IDE's MCP configuration (for VSCode or Cursor, this is
usually an mcp.json file):
{
"mcpServers": {
"biocypher-mcp": {
"url": "https://mcp.biocypher.org/mcp",
"transport": "http"
}
}
}
3. Verify the connection
In your IDE's chat, ask:
The agent should respond with a list of available tools (e.g.
get_cookiecutter_instructions, get_adapter_creation_workflow).
Note
This functionality is currently experimental and does not cover all BioCypher functionality yet. Use with caution.
Next steps
- Agent-supported hands-on tutorial — build a full knowledge graph with the MCP.
- LLM Integration Guide — background and additional LLM-specific documentation.