capa-officer
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- Indirect Prompt Injection (LOW): The skill is designed to ingest and analyze external data such as trigger event documentation and quality data, which creates a surface for indirect prompt injection.\n- Ingestion points: Event documentation and data collection for Root Cause Analysis (RCA) mentioned in SKILL.md.\n- Boundary markers: Absent; there are no instructions to the agent on how to distinguish external data from core instructions.\n- Capability inventory: Mentions several management scripts (e.g., capa-tracker.py, trend-analysis-automation.py) which suggest file or data handling capabilities, although these files were not included in the skill package.\n- Sanitization: No input validation or data sanitization procedures are defined in the provided files.
Audit Metadata