li-prd-review
Pass
Audited by Gen Agent Trust Hub on Sep 24, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process PRD content provided by the user or read from a local file path. This creates a surface for indirect prompt injection where malicious instructions embedded within a PRD could attempt to override the agent's persona, manipulate its responses, or influence how it modifies files.
- Ingestion points: The skill explicitly asks for PRD content or a file path in
SKILL.md(Step 1: 定位 PRD). - Boundary markers: There are no explicit instructions or delimiters used to separate the PRD content from the agent's primary instructions, nor are there warnings to disregard instructions found within the PRD.
- Capability inventory: The agent has the capability to read local files and write updates back to the PRD file (Step 3: 「采纳 → 立即更新 PRD 对应章节」).
- Sanitization: No sanitization, validation, or escaping of the input content is specified in the instructions.
Audit Metadata