mcp-brain-router

mohan-n-swamy/mcp-brain-router

★ 3 stars Python AI/LLM Updated 1d ago

MCP server that routes delegated LLM work across a role-owned provider cascade (GLM, Grok, Codex) with native Claude Code orchestration — quota-only failover, agentic CLI workers.

View on GitHub → 🔍 Audit Wallet Slippage →

Quick Install

Copy the config for your editor. Some servers may need additional setup — check the README.

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "mcp-brain-router": {
      "command": "uvx",
      "args": [
        "mcp-brain-router"
      ]
    }
  }
}

Or install with pip: pip install mcp-brain-router

README Excerpt

An MCP server for Claude Code and Codex that delegates agentic sub-tasks to external CLI workers, returning structured answers with metadata. `mcp-brain-router` is a local Model Context Protocol server shared by Claude Code, Codex, and Grok. Standard calls select a role (`worker`, `simple`, `thinker`, or `adversary`) and run an agentic CLI worker in the caller's absolute working directory. Each role owns an ordered provider shard and advances only on confirmed quota exhaustion. Legacy complexity

Tools (7)

adversarialadversarycheapcodesimplethinkerworker

mcp-brain-router

Quick Install

README Excerpt

Tools (7)

Related AI/LLM Servers