Loop-engineering: 8 unattended agent loop patterns to run your repo while you sleep
JafarNajafov · x · 2026-10-06
A new "loop-engineering" approach: stop prompting and design the loop — agents discover work, do it, verify it, and save state. The repo ships 8 patterns (Daily Triage, PR Babysitter, CI Sweeper, Dependency Sweeper, Changelog Drafter, Post-Merge Cleanup, Issue Triage, Thin loop), one-command setup via npx @cobusgreyling/loop init, and works with Claude Code, Grok, Codex, and Opencode. A loop doctor command scores your "Loop Ready" level weighted by recent runs, and a staged rollout (L1 report-only → L2 assisted → L3 unattended after a week of verifier accuracy) keeps things safe. The repo openly documents failure modes like exploding token costs.
Related event: loop-engineering Repo Goes Viral with 8 Autonomous Agent Loop Patterns(2 posts)→
More from coding & agent
- Laya: Open-Source Model Returns Calibrated Probabilities in 33ms Across 100+ Languages — victormustar · 2026-10-06
- Proactivity-Gym: 23-agent study shows correct work alone can't sustain user trust — minnesotanlp · 2026-10-06
- OpenRUA: off-the-shelf Claude Code drives robots via ROS 2 at 99% success, zero training — Zhaoyang Chu · 2026-10-06
- SearchJev makes search-agent decisions 5x faster with 41-74% lower calibration error — Congfeng Cao · 2026-10-06
- LangChain pitches Managed Deep Agents as a way to monetize personal agent workflows — hwchase17 · 2026-10-06
- Dev envisions local agent 'rooms' no company owns and no server can read — RileyRalmuto · 2026-10-06