photo-privacy-avatar
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill instructions and referenced guides prioritize user privacy and data security. The agent is explicitly instructed to seek authorization for any sensitive file operations and is warned against using external, unauthorized services for processing user photos.
- [INDIRECT_PROMPT_INJECTION]: While the skill processes untrusted user-supplied data (photos, filenames, and documents), it effectively mitigates the risk of indirect prompt injection through explicit negative constraints. 1. Ingestion points: Ingests user-provided image files, filenames, and associated documents as per Section 1 of SKILL.md. 2. Boundary markers: The skill contains explicit instructions in Section 1 to treat text found in images or metadata as data only, ensuring it does not override agent instructions. 3. Capability inventory: The skill utilizes image viewing (view_image), image generation (image_gen), and file system write capabilities. 4. Sanitization: The instructions mandate a clear logical boundary where data processed at runtime is never interpreted as control instructions.
Audit Metadata