afa-scale
Pass
Audited by Gen Agent Trust Hub on Jun 19, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill processes untrusted data in the
user_requestfield which presents a surface for indirect prompt injection. - Ingestion points:
user_requestfield in the context contract (SKILL.md). - Boundary markers: The skill relies on platform-level protocols and a structured output format, though explicit delimiters for user input are not present in the code.
- Capability inventory: Orchestrates routing to other worker skills and maintains state by writing to the
./brand-brain/directory. It does not possess high-privilege capabilities like shell execution. - Sanitization: There is no evidence of specific sanitization for the user-provided request data.
Audit Metadata