from script-kit-next24
Durable pattern and tooling for running bounded autonomous audit loops ("Go AFK for N hours") with cadence, budget, and safe stop conditions.
Looper provides a complete rulebook, templates and scripts to run bounded autonomous audit loops — multi-hour or overnight agent runs that repeatedly inspect, act, and record results. It includes protocol guidance (per-tick prompts), cadence math, budget-edge self-delete rules, templates for scope/stories/logs, and helper scripts to arm and verify a run. The package is opinionated about safety: verifiable stop conditions, cron-friendly minute offsets, and a checklist-based arming protocol so automated loops run predictably.
Use Looper when you need an agent to perform continuous audits or long-running maintenance that must: run unattended for a bounded time, commit verifiable changes to a repo, respect a time budget and automatic shutdown, and keep an auditable trail. Typical triggers: "run overnight", "start an AFK audit for N hours", or "continuously monitor and commit fixes within a budget."
Best for agents that can run shell scripts and cron drivers (Claude Code, local agent runtimes, babashka-style automation). It assumes the agent can create commits, run scheduled ticks, and read/write repo files.
This skill has not been reviewed by our automated audit pipeline yet.
