From loom
Run an unattended / walk-away build session — sets up the verification gate and works a plan to completion autonomously without pausing for questions. Manual-only; invoke with /loom:loom-autopilot.
How this skill is triggered — by the user, by Claude, or both
Slash command
/loom:loom-autopilot [path to plan.md, or a one-line goal][path to plan.md, or a one-line goal]The summary Claude sees in its skill listing — used to decide when to auto-load this skill
You've been asked to run unattended. Drew is not watching. Work the plan to completion, self-correcting via the verification gate, and leave a clear log. Argument: `$ARGUMENTS` (a plan path or a goal).
You've been asked to run unattended. Drew is not watching. Work the plan to completion, self-correcting via the verification gate, and leave a clear log. Argument: $ARGUMENTS (a plan path or a goal).
PLAN.md, or a spec). If none exists, WRITE one first with superpowers:writing-plans — never start implementing without it. There is no human to approve it, so the bar is "a clear, complete plan exists," then proceed (gate #1, no approval pause).mkdir -p .loom && touch .loom/unattended, and ensure .loom/ is in .gitignore. The Stop hook now blocks turn-end until typecheck/build passes..loom/run-<date>.md in Drew's OVERNIGHT_LOG style: Goal, then timestamped actions, decisions, and outcomes.superpowers:subagent-driven-development when tasks are parallelizable./commit (small, well-messaged — the run must be reviewable and rewindable).AskUserQuestion — there is no one to answer. (Loom sets disallowed-tools for this skill as a guard, but treat the rule as binding regardless of whether the runtime enforces it inline.) If you hit a genuinely ambiguous product decision, log it as BLOCKED: , skip that task, and continue with the rest.gh repo create, no dropping data). Log them as DEFERRED for Drew.rm -f .loom/unattended .loom/verify-attempts (disarm the gate and clear the retry counter so the next run starts fresh).superpowers:finishing-a-development-branch to present integration options — but do NOT merge/PR/deploy autonomously (outward-facing; leave it for Drew).Guides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.
npx claudepluginhub mountain-dr3w/loom --plugin loom