claude-council

Why we included this project

Most coding agents sound confident even when they're wrong, and a single model's bias can quietly steer a decision. This plugin runs inside Claude Code and sends the same question to several AI agents at once, showing their answers side by side. It covers Gemini, OpenAI, Grok, Perplexity, and Kimi through their APIs, plus the codex, antigravity, grok, and kimi CLIs if you already have subscriptions, and ollama for fully local use. A synthesis view separates what the agents agree on from where they diverge, and when they do agree it names the assumption that agreement rests on. That makes it a practical cross-check for architecture calls, debugging dead ends, security reviews, and framework choices, especially when the cost of a wrong answer is high.