auditinglisted

Audit a topic in the `m3l-automation` codebase by fanning out parallel read-only Explore agents, aggregating their findings, and then producing a structured plan. **No code, config, or test files are written during this skill** — it ends with an approved plan in plan mode. Findings travel entirely through structured return values, never the filesystem, so this skill is safe to invoke even when plan mode is already active. **This skill must only run in the main (hub) agent, never inside a subagent.** Step 5 calls `EnterPlanMode`, which is only available to the hub. If you find yourself executing this skill as a subagent inside a larger task, stop and surface the audit request back to the hub instead. ## Steps ### 1 — Clarify the target Read the audit target from the user's invocation (e.g. "release config", "scaffolding", "README consistency", "CI pipeline"). If the target is ambiguous or has multiple equally plausible interpretations, ask **one** focused clarifying question before proceeding — otherwise infer reasonable scope from the topic and proceed immediately. Identify 3–5 facets of the topic that a thorough audit should cover. Each facet becomes one Explore agent brief in the next step. Good facets are orthogonal and independently checkable (e.g. for "CI pipeline": step ordering, cache config, gate thresholds, secret handling, trigger conditions). ### 2 — Fan out via the `audit-fanout` workflow Invoke the **`Workflow` tool** with the named workflow `audit-fanout`