loom-auditlisted

# loom-audit Audit is Loom's Ralph-backed adversarial review record surface. It records what was reviewed, what claims or risks were challenged, what context and evidence were inspected, what findings were produced, the auditor's bounded verdict, and what must happen before a consuming surface proceeds. Audit surfaces weak claims, missing or stale evidence, scope drift, and unresolved risk. It does not own intended behavior, ticket closure, risk acceptance, policy, implementation, evidence, or finding disposition. ## Use This Skill When Use this skill when ticket work, records, diffs, evidence, acceptance, closure, research conclusions, specs, plans, constitution changes, packages, or handoffs need independent adversarial review and the result should remain available. Tiny local sanity checks are not audit. Use audit when independent judgment would materially improve trust, acceptance, recovery, or follow-through. ## Ralph Review Requirement Substantive audit requires a bounded Ralph review run. The parent may prepare the request, gather context, and record the result, but the adversarial judgment must come from the Ralph review. Route review from the ticket, evidence, diff, and linked records that define the target. The ticket or audit target carries the durable request and worker contract; the launch prompt is transport. Record the returned judgment in `.loom/audit/`. If Ralph review cannot run, say audit was not performed. Local inspection may still be useful, bu