Agent♥︎Age
Catalog

io.github.qso-graph/qsp-mcp

Official

by qso-graph · Python

QSP — relay MCP tools to any OpenAI-compatible local LLM (llama.cpp, Ollama, vLLM)

QSP — relay MCP tools to any OpenAI-compatible local LLM endpoint.

This server relays MCP tool calls between a local LLM and MCP servers, enabling any model with function calling to access the full qso-graph tool ecosystem (80 tools across 14 packages) from local weights rather than the cloud.

🛠️ Key Features

  • Relays MCP tools to OpenAI-compatible local LLM endpoints
  • Supports local LLMs with function calling (e.g., llama.cpp, Ollama, vLLM)
  • Provides access to the complete qso-graph tool ecosystem
  • No cloud dependency; tools served from local environment

🚀 Use Cases

  • Local LLMs invoking MCP tools for context-aware tasks
  • Offline or privacy-preserving tool access via MCP
  • Integrating 80 tools across 14 packages into local inference workflows

⚡ Developer Benefits

  • Zero-cloud MCP access; run entirely locally
  • Seamless tool calling between LLMs and MCP servers
  • Quick onboarding with install and quick-start guidance
  • Extensible for additional local LLM endpoints

⚠️ Limitations

  • Requires a compatible local LLM with function calling support
  • May rely on local environment setup for MCP server connectivity
  • Documentation references indicate installation via pip and config setup; ensure environment paths are correct