writing-plans
Pass
Audited by Gen Agent Trust Hub on Jul 14, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection vulnerability surface by processing external requirements to generate implementation tasks.
- Ingestion points: Untrusted data enters the context via the 'spec or requirements' provided by the user for multi-step tasks.
- Boundary markers: The skill lacks explicit boundary markers or instructions to the agent to disregard embedded instructions within the user-provided specs when generating the plan.
- Capability inventory: The generated plans involve sensitive operations including file system writes (creating and modifying source code) and shell command execution (
pytest,git commit) when processed by follow-up execution skills. - Sanitization: There is no mechanism described for sanitizing or validating user input before it is interpolated into the executable implementation steps.
Audit Metadata