anthropic-art
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied topics and text to construct prompts for an image generation tool.
- Ingestion points: User-specified topics, metaphors, and verbatim text fields in
SKILL.md. - Boundary markers: Instructions define a strict "Prompt contract" template to constrain the output format.
- Capability inventory: Uses a built-in image generation tool to create bitmaps.
- Sanitization: Employs a "Quality gate" that instructs the agent to reject or revise outputs that deviate from specific style and safety requirements.
- [EXTERNAL_DOWNLOADS]: The skill's style specification includes links to official Anthropic visual assets hosted on a well-known content management service.
- Evidence: References to
https://cdn.sanity.io/images/...andhttps://www.anthropic.com/newsinreferences/style-spec.md.
Audit Metadata