task-breakdown
Pass
Audited by Gen Agent Trust Hub on Jul 12, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill structure and instructions align perfectly with its stated purpose of assisting in development planning. No hidden malicious commands or exfiltration patterns were found.
- [SAFE]: Indirect Prompt Injection Surface analysis: (1) Ingestion point: User-provided Design Docs. (2) Boundary markers: Present in subagent prompt templates using bold headers and delimited sections. (3) Capability inventory: Uses standard file reading and subagent orchestration tools. (4) Sanitization: Relies on LLM instruction following within templates. The implementation is considered safe for standard development environments.
Audit Metadata