Graph Tool Call

Example Use Cases

• 1

  Retrieve relevant tools with workflow context rather than isolated matches.

• 2

  Reduce token usage by filtering and chaining tools intelligently.

• 3

  Integrate multiple MCP servers behind a single proxy to simplify tool access.

Description

graph-tool-call constructs a tool graph from various sources like OpenAPI, MCP, and Python functions to enable workflow-aware retrieval of multi-step tool chains. It improves accuracy and reduces token usage significantly compared to baseline vector search by combining BM25, graph traversal, embedding similarity, and MCP annotations. The server supports integration with multiple MCP-compatible agents and can proxy multiple MCP servers to reduce token overhead.