pm-feature
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill analyzes external documents such as product requirements and design prototypes to generate detailed functional specifications, creating a surface for potential instruction injection via data content.
- Ingestion points: According to Steps 2 and 3 in
SKILL.md, the skill reads existing documentation such asdocs/02-方案设计/PRD产品需求文档.mdand原型设计方案.md. - Boundary markers: There are no explicit delimiters or instructions directing the agent to ignore potential commands embedded within the analyzed documents.
- Capability inventory: The skill utilizes
Bashfor file status checks,Writefor generating task documents, and theAgenttool for delegating sub-tasks to subagents. - Sanitization: The skill does not perform validation, sanitization, or filtering of the text extracted from input documents before summary or re-writing.
Audit Metadata