brainstorm-mcp

Overview

brainstorm-mcp is a sophisticated MCP server that orchestrates debates between multiple AI models including GPT, Gemini, DeepSeek, Claude, and Ollama. It operates in two modes: API mode for direct calls to external providers, and hosted mode that uses models already available in the user's environment. The server enables structured multi-round debates where models critique each other's responses, followed by synthesis that provides a 3-bullet verdict with recommendation, key tradeoffs, and strongest disagreement. Features include parallel execution for quick insights, error resilience when one model fails, and support for different debate styles including freeform, red-team (adversarial), and Socratic approaches.

Try asking AI

After installing, here are 5 things you can ask your AI assistant:

• brainstorm

  Multi-round debate between AI models (API or hosted mode)

  Try:Brainstorm using opus, sonnet, and haiku about whether we should use GraphQL or RESTTry:Debate the merits of microservices vs monolithic architecture
• brainstorm_quick

  Instant multi-model perspectives — parallel, no rounds

  Try:Use brainstorm_quick to compare Redis vs PostgreSQL for session storageTry:Quick comparison of Python vs Go for backend development
• brainstorm_review

  Multi-model code review with findings, severity, verdict

  Try:Review this diff for security issuesTry:Analyze this pull request for performance improvements
• brainstorm_respond

  Submit Claude's response in an interactive session

  Try:Provide my response to the current debate roundTry:Submit my analysis to the multi-model discussion
• brainstorm_collect

  Submit model responses in a hosted session

  Try:Collect responses from all models in this sessionTry:Gather the latest model responses for synthesis
• list_providers

  Show configured providers and API key status

  Try:Show me which AI providers are configuredTry:List all available AI models and their status
• add_provider

  Add a new AI provider at runtime

  Try:Add Groq as a new provider with model llama3-8bTry:Configure OpenAI with GPT-4o model

Comparable tools

Installation

{
  "mcpServers": {
    "brainstorm": {
      "command": "npx",
      "args": ["-y", "brainstorm-mcp"],
      "env": {
        "OPENAI_API_KEY": "sk-...",
        "GEMINI_API_KEY": "AIza...",
        "DEEPSEEK_API_KEY": "sk-..."
      }
    }
  }
}

{
  "mcpServers": {
    "brainstorm": {
      "command": "npx",
      "args": ["-y", "brainstorm-mcp"],
      "env": {
        "OPENAI_API_KEY": "sk-...",
        "DEEPSEEK_API_KEY": "sk-..."
      }
    }
  }
}

npm install -g brainstorm-mcp
brainstorm-mcp

FAQ

Does this require API keys to work?

No, it has a hosted mode that works without API keys by using models already available in your environment (Claude Opus/Sonnet/Haiku via sub-agents). API mode requires keys for external providers.

Can I use custom models not listed in the documentation?

Yes, you can add custom providers at runtime using the `add_provider` tool or through the configuration file by specifying a custom baseURL and model name.

On Hacker News

Recent discussion from the developer community.

• Show HN: Brainstorm-MCP – Let GPT, DeepSeek, and Groq Brainstorm Together▲ 2💬 1
  Story by spranab · 2026-02-24

Compare brainstorm-mcp with