investigate-last-crash

Pass

Audited by Gen Agent Trust Hub on Jun 17, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes a series of standard Linux diagnostic utilities including journalctl, coredumpctl, sar, and ls to retrieve system state and logs.
  • [DATA_EXPOSURE]: The skill reads sensitive system information including kernel ring buffers and system logs. This behavior is entirely consistent with the primary purpose of a crash investigation tool and the results are stored locally in the user's home directory (~/Daniel-Workstation-Updates/).
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from system logs (journalctl, coredumpctl) which could theoretically contain malicious instructions if a local process or remote service logged crafted strings.
  • Ingestion points: System logs via journalctl and coredumpctl (SKILL.md).
  • Boundary markers: Absent; log content is directly used to generate the report.
  • Capability inventory: File write access to the local filesystem (SKILL.md).
  • Sanitization: Absent; the skill extracts and quotes log lines without validation.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 17, 2026, 10:52 AM
Security Audit — agent-trust-hub — investigate-last-crash