Panel MCP Server

An MCP (Model Context Protocol) server that provides LLM council workflow tools for multi-model queries, debates, and reviews.

Overview

Panel MCP Server enables AI clients (like Claude) to orchestrate multi-model workflows. The calling LLM uses these tools to query multiple models in parallel, run structured debates, get critiques, and then synthesize the results itself.

Key insight: The MCP server exposes tools that return structured multi-model responses. The calling LLM decides when/how to use these tools and can further process, compare, or synthesize the collected responses using its own reasoning.

Features

• council_query: Query multiple LLMs in parallel, returning all responses for synthesis
• debate: Run structured adversarial debates between two models
• critique: Get one model to critique another's response
• query_model: Query a single model directly
• list_models: Discover available models by provider
• health_check: Check provider status and connectivity

Provider Support

Dual-Mode Provider System

1. OpenRouter mode (recommended): Use openrouter/ prefix for any of 300+ models

   • openrouter/anthropic/claude-sonnet-4
   • openrouter/openai/gpt-4o
   • openrouter/meta-llama/llama-3.3-70b-instruct

2. Direct mode: Use provider prefix for direct API calls (lower latency, no fee)

   • openai/gpt-4o - calls OpenAI API directly
   • anthropic/claude-sonnet-4-20250514 - calls Anthropic API directly
   • google/gemini-2.5-pro - calls Google API directly

Default Panel: GPT-4o + Claude Sonnet 4 + Gemini 2.5 Pro

Installation

npm install panel-mcp-server
# or
pnpm add panel-mcp-server

Configuration

Set at least one API key:

# Required (at least one)
export OPENROUTER_API_KEY=sk-or-...

# Optional direct providers (lower latency)
export OPENAI_API_KEY=sk-...
export ANTHROPIC_API_KEY=sk-ant-...
export GOOGLE_GENERATIVE_AI_API_KEY=...
export MISTRAL_API_KEY=...

# Optional configuration
export PANEL_DEFAULT_MODELS=gpt-4o,claude-sonnet-4-20250514,gemini-2.5-pro
export PANEL_MAX_CONCURRENT=5
export PANEL_REQUEST_TIMEOUT_MS=60000

Usage

As MCP Server (stdio)

panel-mcp-server --stdio

As MCP Server (HTTP)

panel-mcp-server --http 8080

Claude Desktop Configuration

Add to your Claude Desktop config (~/Library/Application Support/Claude/claude_desktop_config.json):

{
  "mcpServers": {
    "panel": {
      "command": "npx",
      "args": ["panel-mcp-server", "--stdio"],
      "env": {
        "OPENROUTER_API_KEY": "your-key-here"
      }
    }
  }
}

Tools

council_query

Query multiple LLM models in parallel. Returns all responses for synthesis.

{
  prompt: string,
  models?: string[],  // defaults to GPT-4o, Claude Sonnet 4, Gemini 2.5 Pro
  systemPrompt?: string
}

debate

Run a structured adversarial debate between two models.

{
  topic: string,
  affirmativeModel: string,
  negativeModel: string,
  rounds?: number  // 1-5, default: 2
}

critique

Have one model critique a response.

{
  originalPrompt: string,
  response: string,
  criticModel: string,
  aspects?: string[]  // e.g., ["accuracy", "completeness"]
}

query_model

Query a single model directly.

{
  prompt: string,
  model: string,
  systemPrompt?: string
}

list_models

List available models by provider.

{
  provider?: "all" | "openrouter" | "openai" | "anthropic" | "google" | "mistral"
}

health_check

Check provider status.

{
}

Development

# Install dependencies
pnpm install

# Validate (format + lint + test + build)
pnpm validate

# Development mode
pnpm dev

# Run tests
pnpm test

Built With

• FastMCP - MCP server framework
• Vercel AI SDK - LLM provider integrations
• functype - Functional programming patterns
• ts-builds - TypeScript build tooling

License

MIT