Hi Product Hunt, I built Intent CLI because AI coding agents are getting very good at writing code, but real product development still breaks when the intent is scattered across prompts, chats, issues, and memory. The core idea is simple: before asking AI to implement, capture the product intent. Keep that intent in a versioned tree. Then let agents turn it into focused GitHub issues, implement one slice at a time, review against the original intent, and recover when the loop gets stuck. Intent CLI is not another code generator. It is a deterministic workflow layer for Claude, Codex, and other AI coding agents. The human talks naturally in a design thread, while the agent asks intent-cli for the current guidance, status, packets, review rules, and recovery path. What it helps with: - turning product intent into structured implementation work - keeping design decisions and ADRs visible over time - creating GitHub issues from intent-backed packets - running implementation and review loops with AI agents - detecting drift, repeated failure loops, and missing clarifications - recovering when labels, reviews, or metadata get out of sync It is open source and built as a .NET CLI. The long-term goal is to make AI-assisted development less prompt-driven and more intent-driven. I would love feedback from people using AI coding agents in real projects: Where does your current AI development workflow lose context? What part of design, issue creation, review, or recovery should be more structured?

Intent CLI

Intent-driven development for AI coding agents

Intent-driven development for AI coding agents