Council – Run Claude, Codex and Gemini against the same prompt

Name: Council – Run Claude, Codex and Gemini against the same prompt
Availability: InStock
Author: colinarms

by colinarms·May 1, 2026·2 points·0 comments

Visit Project View on HN

AI Analysis

●●●BangerBig BrainSolve My Problem

Surfaces model disagreements instead of averaging them away — that's the real value.

Strengths

•Parallel execution with heartbeat monitoring prevents silent failures on slow models.
•Pipe-friendly design integrates into CI pipelines and Node scripts via --json flag.
•Uses existing subscriptions — no new API keys or vendor lock-in required.

Weaknesses

•Limited to three models; no extensibility for open-source or self-hosted LLMs yet.
•Synthesizer logic is opaque — unclear how disagreements are weighted or resolved.

Post Description

I often copy and paste the same prompts into Claude, Codex & Gemini separately. It's helpful seeing where they all agreed and where they diverged. The signal of the agreement or divergence was often more helpful than any individual response.

So, I built a tiny open source CLI that does this automatically.

Council detects any `claude`, `codex` or `gemini` on your path and runs them in parallel. Once all finish, a lead summarizer produces one synthesized answer, flagging any disagreements.

Open source & MIT licensed: https://github.com/seeARMS/council