system-scan-agentlisted

# Linux system scan with independent review Use this skill for the “adapters + AI triage + independent review” branch of the unified workflow. Follow bootstrap, toolchain validation, local/remote target selection, and host-AI triage first. Once triage is recorded, ask a separate question: “May I share the bounded, redacted results and AI triage with the independent system-triage agent?” Do not delegate before a clear yes. The host-AI step may use `trustedAi` only after its own separate approval; that expands non-secret diagnostics but never permits credentials or keys. The reviewer is not another scanner: it validates only the supplied evidence, distinguishes confirmed facts from hypotheses, and does not modify the host. It may call `system_advisory_lookup` only after a distinct approval for that network request; otherwise it marks advisory claims unverified. Record its exact response with `system_record_run` using `agent_review`, then finalize in `scan-agent` mode. Keep scanner facts, host-AI notes, and reviewer notes as separate sections.