investigate-bug
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input during the investigation process. Ingestion points: User-provided arguments ($ARGUMENTS) and gathered environment symptoms (SKILL.md). Boundary markers: None present to isolate untrusted data from the agent's instructions. Capability inventory: Explicitly invokes a 'consultant CLI' (SKILL.md). Sanitization: No evidence of input validation or sanitization for external symptoms.
- [COMMAND_EXECUTION]: The skill is configured to invoke a 'consultant CLI' to analyze system state and root causes.
- [NO_CODE]: The skill contains only documentation and instructional content in SKILL.md with no additional executable code or scripts.
Audit Metadata