REAP - Recursive Evolutionary Autonomous Pipeline

Hey Product Hunt! 👋

I'm the maker of REAP. I built this because I kept hitting the same wall when coding with AI agents like Claude Code:

Every new session, the AI forgets everything. The architecture decisions we made yesterday, the lessons from that painful debugging session, the conventions we agreed on — all gone. I'd spend the first 10 minutes of every session re-explaining context, and the AI would still make the same mistakes we'd already fixed.

So I asked: What if a project could have DNA — a living memory that evolves as the codebase grows?

That's REAP. It introduces a Genome (architecture principles, business rules, conventions) that persists across sessions and evolves through generations. Each generation follows a structured lifecycle:

Objective → Planning → Implementation → Validation → Completion

The key insight: when the AI discovers a design flaw during implementation, it doesn't just hack a fix. It logs the issue in a backlog, and at Completion, the Genome itself evolves. Next generation, the AI starts smarter.

What makes REAP different:
- 🧬 Your project's "DNA" accumulates wisdom across every session
- 🔄 Generation-based workflow prevents scattered, context-free coding
- 📋 Backlog system ensures nothing gets lost between sessions
- 🪝 SessionStart hooks auto-inject full project context — zero re-explaining
- 🛠️ Works with Claude Code today, designed to be agent-agnostic

REAP is fully open-source (MIT) and dogfoods itself — REAP is built using REAP.

npm install -g @c-d-cc/reap
reap init

I'd love to hear how you manage context and continuity when working with AI coding agents. What's your biggest pain point while vibe coding?