Practical patterns, starters & CLI tools for loop engineering with AI coding agents. Design systems that prompt and orchestrate agents (inspired by Addy Osmani and Boris Cherny). Includes loop-audit, loop-init, loop-cost.
TypeScript agentic-aiai-agentsai-codinganthropicautomationclaudeclaude-codecodexcoding-agentsdevops-automationdevtoolsgithub-actionsgrokllmloop-engineeringmcpprompt-engineering
AGENTS.md — Dependency Sweeper (Opencode)
Loop Mode
- L2 assisted from week one (patch-only with strong verifier gates).
- Read
dependency-sweeper-state.mdbefore every run.
Safety
- Patch-only by default. Majors and breaking changes escalate to human.
- Maintain denylist in state file — never touch those packages.
- One worktree per update attempt.
- Max 3 fix attempts per run; max 5 auto-PRs per day.
Verification
- Verifier must run
npm ci && npm test(or equivalent). - Record attempt count and test evidence in state.