muapi-instagram-post
Pass
Audited by Gen Agent Trust Hub on Sep 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user data from the
briefandbrand_styleinputs, which are interpolated directly into instructions for the LLM (e.g., in Phase A, Step 1). There is a lack of boundary markers (such as XML tags or delimiters) or explicit instructions to ignore embedded commands within these inputs, creating a surface for indirect prompt injection where a user could potentially override the skill's intended behavior. - Ingestion points:
briefandbrand_styleinputs inSKILL.md. - Boundary markers: None present.
- Capability inventory: Execution of
muapiCLI commands andcurlrequests to external APIs. - Sanitization: No input validation or escaping is performed before interpolation.
Audit Metadata