brainstorm-mcp and mentor-mcp-server
These are **complements** — the brainstorm server orchestrates multi-model debates to generate diverse reasoning perspectives, while the mentor server provides structured deep reasoning critique on those outputs through Deepseek R1's chain-of-thought capabilities, enabling iterative refinement workflows.
About brainstorm-mcp
spranab/brainstorm-mcp
MCP server for multi-round AI brainstorming debates between multiple models (GPT, DeepSeek, Groq, Ollama, etc.)
Operates in dual modes—API mode for direct parallel calls to external providers, or hosted mode using Claude sub-agents with zero API key setup. Structures debates across multiple rounds with models critiquing each other's responses, then synthesizes findings into 3-bullet verdicts (recommendation, tradeoffs, disagreements). Includes specialized tools for code review with severity ratings and confidence scoring based on model agreement, plus a quick mode for instant parallel perspectives without debate rounds.
About mentor-mcp-server
cyanheads/mentor-mcp-server
A Model Context Protocol server providing LLM Agents a second opinion via AI-powered Deepseek-Reasoning R1 mentorship capabilities, including code review, design critique, writing feedback, and idea brainstorming through the Deepseek API.
Related comparisons
Scores updated daily from GitHub, PyPI, and npm data. How scores work