nizos

nizos / conduct

Public

Process discipline for AI coding agents

claude-code codex coding-agent github-copilot guardrails
29
2
100% credibility
Found May 01, 2026 at 29 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
TypeScript
AI Summary

Conduct enforces customizable rules on AI coding agents to promote disciplined development practices like test-driven development and pattern restrictions.

How It Works

1
📰 Discover a helpful guide

You hear about Conduct from a friend or online, a simple way to make AI coding helpers follow your rules for better code.

2
📥 Bring it into your project

You easily add it to your coding workspace with one quick step.

3
⚙️ Choose your rules

You pick friendly rules like 'write tests first' or 'use consistent names' to keep everything neat.

4
🔗 Link to your AI helper

You connect it so your AI coding buddy always checks the rules before making changes.

5
👀 See it in action

As you chat with your AI and it suggests code, the rules gently block messy ideas and guide good ones.

🎉 Enjoy clean, smart coding

Your project stays organized and strong, with AI helping perfectly every time.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 29 to 29 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 conduct?

Conduct is a TypeScript CLI tool that enforces process discipline for AI coding agents like Claude Code, GitHub Copilot, and OpenAI Codex. It intercepts file writes and shell commands from agent hooks, blocking violations of your rules—like requiring a failing test before code (TDD enforcement), filename casing standards, or forbidden patterns such as "npm install" in a pnpm setup. Install as a dev dep, configure rules in conduct.config.ts, and pipe payloads: `conduct --agent claude-code`.

Why is it gaining traction?

It generalizes single-agent guards into multi-vendor support with deterministic rules (pattern forbids) and AI validators for nuanced checks like TDD. Devs get instant workflow wins: no more rogue agent commands causing github process completed with exit code 1, or undisciplined writes skipping tests. Custom rules extend it cheaply, making it a flexible conductor for AI outputs without rebuilding agent integrations.

Who should use this?

Teams using AI agents for coding who need a discipline process in the workplace, like enforcing TDD on src/** paths or banning emojis in docs. GitHub Copilot users tired of unchecked bash runs, or devs maintaining a code of conduct github template across agents. Perfect for disciplined process synonyms in collaborative repos.

Verdict

Early with 29 stars and 1.0% credibility score, but solid docs, tests, and MIT license make it low-risk to try. Adopt if AI agents need conducting; skip for solo scripting.

(187 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.