loop

A loop is not a saved artifact like a flow. It is a tracking doc plus a run: /loop writes .codevoyant/loops/{slug}/loop.md (task, objective, bound, and a per-iteration log) and immediately executes iterations, appending each result to the doc, until the objective is met or the bound is reached. There is nothing to define ahead of time and no separate run command.

Markdown output: soft-wrap prose, never hard-wrap — when this skill writes a .md artifact (the tracking doc or any generated document), write each paragraph as one continuous line; do not insert manual newlines to wrap prose at a fixed column width. Newlines still separate paragraphs, list items, headings, and code fences.

Usage

/loop <task to repeat> --until <objective> [--max N] [--check <command>] [--resume <slug>]

• task (required positional) — what to repeat each iteration: a skill command, shell command, or agent instruction. Runs in a background agent.
• --until (required) — the objective: the verifiable condition that ends the loop, phrased as an outcome, not an activity.
• --max N (default 3, must be ≥ 1) — the hard upper bound. The loop stops after N iterations even if the objective is not met.
• --check (optional) — a deterministic check that exits 0 when the objective is met. When present it is the authoritative signal and overrides the agent's verdict.
• --resume (optional) — continue an existing tracking doc (append iterations under its existing task/objective/max) instead of starting a new one.

Procedure