goose-ads
Pass
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data which could contain malicious instructions.
- Ingestion points: Brand knowledge actions, external playbooks (via catalog_fetch), and user-uploaded files (media_upload) as described in SKILL.md.
- Boundary markers: Relies on instructions to follow 'active playbooks' and 'required gates' to structure agent behavior.
- Capability inventory: Includes tools for reading/updating creative assets, resizing, and fetching additional skills.
- Sanitization: Implements safety rules that explicitly forbid the agent from requesting keys, tokens, or passwords, reducing the risk of data exfiltration attacks.
Audit Metadata