onevcat

onevcat / argue

Public

Harness-agnostic orchestration package for multi-agent consensus workflows

18
3
100% credibility
Found Apr 11, 2026 at 18 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
TypeScript
AI Summary

A structured debate system where multiple AI agents independently analyze problems, critique each other over rounds, vote to consensus, and produce reliable outputs with optional follow-up actions.

How It Works

1
🔍 Discover argue

You learn about a clever tool that makes AI give smarter answers by having multiple AIs debate like friends in a group chat.

2
đź“Ą Pick it up

You grab the argue tool and add it to your computer in a quick step.

3
🤝 Team up AIs

You choose a couple of your favorite AI helpers and link them so they can chat together.

4
đź’¬ Ask a big question

You share a tricky problem, like deciding on the best way to organize project files, and launch the debate.

5
đź‘€ Watch the fun

You follow along as the AIs propose ideas, challenge each other, vote, and reach a smart agreement.

🎉 Celebrate the answer

You get a clear summary of the winning ideas, backed by their agreement, ready to guide your next move.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 18 to 18 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is argue?

Argue is a TypeScript package and CLI for harness-agnostic orchestration of multi-agent consensus workflows. You feed it a task like "Should we use monorepo or polyrepo?", and multiple AI agents independently propose claims, debate them across rounds—challenging, revising, merging—and vote to reach consensus. Users get structured outputs: surviving claims with vote tallies, peer-reviewed scores, a final report, and optional actions like opening a GitHub PR, all while cutting hallucinations through enforced argue culture.

Why is it gaining traction?

Unlike single-model chats or brittle agent chains, argue enforces rigorous debate—agents must defend claims, self-correct, and quantify agreement—yielding higher-quality results than solo agents. The CLI spins up debates in one command with providers like Claude or OpenAI, spitting out JSONL logs, results, and Markdown summaries for easy auditing. Its library mode hooks into any agent framework via a simple delegate, making multi-agent orchestration dead simple without vendor lock-in.

Who should use this?

AI workflow builders scripting agent debates for code reviews, architecture decisions, or issue triage. DevOps teams automating PR fixes via consensus-voted actions. Researchers prototyping argued consensus in multi-agent systems, especially those juggling CLI tools like Codex or custom SDKs.

Verdict

Grab the CLI for quick, structured agent debates—docs are excellent in English, Chinese, and Japanese, with real examples like auto-fixing issues. At 18 stars and 1.0% credibility, it's early-stage (light tests, nascent community), so prototype cautiously, but the harness-agnostic design nails multi-agent workflows today.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.