Skip to main content
MCP server CX Codex

brainstorm-mcp — Claude Code MCP server: what it does, how to install

Runs multi-model debates inside your editor to stress-test architectural choices and code changes across competing AI perspectives.

Install into Claude Code · Platform CLI All methods and platforms ↓
claude mcp add brainstorm-mcp -e OPENAI_API_KEY=<OPENAI_API_KEY> -e GEMINI_API_KEY=<GEMINI_API_KEY> -e DEEPSEEK_API_KEY=<DEEPSEEK_API_KEY> -- npx -y brainstorm-mcp
Open on GitHub

What brainstorm-mcp does

The server orchestrates multi-round debates between different LLMs instead of relying on a single answer. Models inspect responses from rivals and challenge weak spots across rounds. At the end, the server returns a three-point verdict covering the recommendation, key tradeoffs, and the core disagreement.

Execution splits across hosted, API, and CLI channels. Hosted mode operates without separate API keys by querying models within the host setup. Supplying keys like OPENAI_API_KEY or GEMINI_API_KEY triggers direct concurrent queries to external endpoints, including local Ollama instances. It also discovers command-line agents on your PATH to run queries through subscription quotas instead of API credits.

Tools include brainstorm_review for structured code audits and brainstorm_quick for rapid multi-angle feedback under ten seconds. Without external API credentials or installed agent CLIs, debates stay constrained to the host environment.

What it helps with

  • compare competing model assessments before finalizing architecture decisions
  • run a multi-model code review with severity ratings using brainstorm_review
  • challenge implementation plans through red-team debate styles
  • gather fast alternative perspectives under ten seconds with brainstorm_quick

Example prompts for the agent

  • Debate whether to adopt REST or GraphQL using sonnet and opus.
  • Run brainstorm_review on this pull request diff using red-team mode.
  • Get fast multi-model feedback on this database schema with brainstorm_quick.

How to install brainstorm-mcp

Platform CLI

claude mcp add brainstorm-mcp -e OPENAI_API_KEY=<OPENAI_API_KEY> -e GEMINI_API_KEY=<GEMINI_API_KEY> -e DEEPSEEK_API_KEY=<DEEPSEEK_API_KEY> -- npx -y brainstorm-mcp

Replace <…> with your own values

Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more

Questions about brainstorm-mcp

How do I install brainstorm-mcp in Claude Code?

Run `claude mcp add brainstorm-mcp -e OPENAI_API_KEY=<OPENAI_API_KEY> -e GEMINI_API_KEY=<GEMINI_API_KEY> -e DEEPSEEK_API_KEY=<DEEPSEEK_API_KEY> -- npx -y brainstorm-mcp`. Provider keys can be omitted when running solely in hosted mode.

What does brainstorm-mcp do?

It pits multiple language models against each other in structured cross-examination debates. Instead of trusting a single prediction, you see opposing critiques and a distilled summary of disagreements.

Are paid API keys required to run brainstorm-mcp?

Separate API keys are necessary only for querying third-party provider endpoints directly. Hosted mode relies on the host client, while CLI mode executes through local agent binaries already tied to your subscriptions.

Similar MCP servers

All MCP servers for Agent setup: memory, context, orchestration →

Project sponsors

Companies that keep this analytics open
Gold sponsor

Impressions, clicks and code copies — open statistics. We keep the numbers public for the same reason NeBlask itself is public. Become a sponsor