creative-director
Pass
Audited by Gen Agent Trust Hub on Aug 11, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted external data via the
[PASTE ANGLES]placeholder. If the provided angles contain malicious instructions, the agent could potentially be manipulated into performing unintended actions. This is particularly relevant as the skill has write access to the Meta Ads platform through an MCP. The risk is mitigated by a mandatory manual review step where the user must approve the generated images before they are uploaded. \n - Ingestion points: External data enters the context via the
[PASTE ANGLES]block inSKILL.md. \n - Boundary markers: The instructions lack explicit delimiters or instructions to ignore embedded commands within the input data. \n
- Capability inventory: The skill uses an MCP with
writeaccess to upload images to an ad library and provides image hashes for subsequent campaign building. \n - Sanitization: No filtering or validation of the input 'angles' is defined.
Audit Metadata