multi-llm-mcp-server

Skill by ara.so — MCP Skills collection.

An MCP server that enables Claude Code to delegate tasks to Codex CLI and query multiple LLM providers (GPT, Kimi, DeepSeek, Qwen, Claude) in parallel or individually. Built on FastMCP, it solves two key problems: (1) letting Claude Code offload work to Codex CLI, and (2) querying multiple models simultaneously for comparison and cross-referencing.

Installation

Prerequisites

# Python 3.10+ required
pip install fastmcp openai

# Optional: Install Codex CLI for ask_codex functionality
# Ensure 'codex' is available in PATH and authenticated

Environment Variables