requirements-from-decisions
Pass
Audited by Gen Agent Trust Hub on May 18, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from local project files which could potentially contain malicious instructions intended to influence the agent's behavior. However, the skill restricts its operations to drafting structured markdown and does not possess high-risk capabilities like network access or arbitrary command execution.
- Ingestion points:
.agents/project-context.mdand.agents/decisions.mdin theSKILL.mdworkflow steps. - Boundary markers: Absent; the skill directly incorporates content from these files into the requirement drafting process.
- Capability inventory: File system write access limited to
.agents/requirements.md. - Sanitization: The skill relies on specific structural instructions (e.g., table formats, status legends) and LLM safety filters to maintain output integrity.
Audit Metadata