council-mcp-server: Instructions file for Claude Code

CLAUDE.md - Council MCP Server

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

@~/.claude/CLAUDE.md

Project Overview

Council is a Model Context Protocol (MCP) server that enables Claude to collaborate with multiple AI models via OpenRouter. It provides a provider-agnostic way for AI-to-AI collaboration on complex tasks, allowing access to Gemini, GPT, Claude, Llama, and many other models.

Key Features

• Multi-Model Support: Access 100+ models via OpenRouter (Gemini, GPT, Claude, Llama, Mistral, etc.)
• Dynamic Model Discovery: List and filter available models by provider, capability, or pricing
• Per-Request Model Override: Use different models for different tasks
• Multiple Collaboration Tools: Code review, brainstorming, test generation, explanations
• Response Caching: Automatic caching for repeated queries

Available MCP Tools

Since this MCP server is already running, you can use these tools directly:

Core Tools

• mcp__council__ask - Ask any model general questions
• mcp__council__code_review - Get code review
• mcp__council__brainstorm - Brainstorm ideas with AI
• mcp__council__test_cases - Generate test cases
• mcp__council__explain - Get explanations
• mcp__council__synthesize_perspectives - Combine multiple viewpoints

Model Management

• mcp__council__server_info - Check server status and current model
• mcp__council__list_models - List available models with filtering
• mcp__council__set_model - Change the active model

Using Model Override

All tools support an optional model parameter to override the default model:

# Use a specific model for a code review
mcp__council__code_review(
    code="def hello(): print('world')",
    focus="security",
    model="anthropic/claude-3-opus"  # Override default model
)

# Ask a specific model
mcp__council__ask(
    question="Explain quantum computing",
    model="google/gemini-3-pro-preview"
)