The original title is: "I Wrote a Claude Skill That Sets Up Loop Engineering in Your Repo"

Short summary

Loop engineering replaces manual back-and-forth prompting with an autonomous two-agent system: a maker handles tasks, an independent checker verifies work, and a persistent state file (LOOP.md) manages memory. A new Claude Code skill sets this up in one command. Human gates remain on all irreversible actions—merging, deploying, deleting, spending.

• •Autonomous loop: maker agent + independent checker + persistent LOOP.md state file
• •One-command setup scaffolds everything into .claude/agents/
• •Human gates protect all irreversible actions (merge, deploy, delete, spend)

Generated with AI, which can make mistakes.