system-scan-ailisted

# Linux system scan with AI triage Use this skill for the “adapters + AI triage” branch selected from the unified workflow. Follow `system-scan` through bootstrap, toolchain validation, plan, preview, and per-adapter consent first. Ask separately whether redacted findings may be shared with the host model; no model analysis occurs until the answer is an unambiguous yes. Ask separately whether the selected AI is trusted to receive expanded non-secret diagnostics (`trustedAi`). Pass that flag only when approved. After the evidence is complete, call `system_ai_triage_payload` in batches of 40 using `findingOffset` when needed. Record each batch with `host_ai_triage` and the same offset; the server merges batches and rejects missing indexes at finalization. The model must classify every finding as `true_positive`, `false_positive`, or `needs_review`, cite only the provided evidence, and treat package CVE matches as unverified until distribution backports are checked. Then finalize in `scan-ai` mode. AI judgement is advisory. Even in `trustedAi` mode, do not give it raw packet captures, credentials, tokens, private keys, authentication headers, or unredacted scanner reports.