problem-analysis
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- PROMPT_INJECTION (LOW): SKILL.md and CLAUDE.md contain instructions to IMMEDIATELY invoke the script and Do NOT explore first, which attempts to override the agent's default safety analysis of tool parameters.
- COMMAND_EXECUTION (LOW): The skill runs a local Python script (skills.problem_analysis.analyze). The README notes that the script dynamically generates commands for the agent to run. The script source is not provided in the analyzed files.
- INDIRECT_PROMPT_INJECTION (LOW): The skill processes untrusted user-provided problem reports during investigation. Evidence Chain: 1. Ingestion points: Problem descriptions in README.md usage. 2. Boundary markers: Absent. 3. Capability inventory: Python script execution and dynamic command generation. 4. Sanitization: None documented.
Audit Metadata