michi-looplisted

# Michi Loop Run a task as an autonomous, lights-off **loop** — the agent works unattended, checks its own output against a verifier, and stops on a defined condition — *with Michi discipline around it.* This skill is **discipline over mechanism**. It does not run the loop; Claude Code already ships the runners (`/loop`, `/goal`, `claude -p`, `Workflow`). What it adds is the judgment the loop-engineering literature skips: *have you earned lights-off, what does "done" mean, what checks the work, and when does it stop?* You are invoked once, as the **designer and launcher** of a loop — not its body. It is the **convergent + autonomous** lane — the sibling of `michi-expedition` (emergent + Entrusted-spiral). The boundary between them is the *form of verification you can answer* (see the Gate). ## When to use (and when not) **Use it when** the work is *convergent* (you can state "done when ___"), the verifier is *trustworthy* (a real check, not the agent's say-so), the blast radius is *reversible* (a container, a worktree, a branch), and autonomy has been *earned* (you've watched this kind of work go well Paired). **Don't** when the work is exploratory or the end isn't knowable — that's `michi-expedition`. Don't when you can't answer "done when ___". Don't when the only verifier is human visual judgment with no automatable floor — a loop can't see the screen. **Posture: surface-and-explore, you decide.** This skill *raises* the readiness checks; it does not police them. Th