ha-frontend-review
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to dynamically ingest instructions from external companion files.
- Ingestion points: The 'Review Preparation' section in
SKILL.mddirects the agent to use the../../../AGENTS.mdfile to identify and load matching companionSKILL.mdfiles. - Boundary markers: The instructions do not provide explicit delimiters or "ignore" directives to protect the agent from adversarial instructions that might be present in the loaded companion files.
- Capability inventory: The skill primary capability is generating natural language review feedback and PR body content. No dangerous system-level capabilities (like shell execution or network writes) are defined in this specific skill.
- Sanitization: There is no evidence of validation or sanitization of the content fetched from the referenced companion skills.
Audit Metadata