image-to-motion
Pass
Audited by Gen Agent Trust Hub on Oct 24, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill architecture relies on processing untrusted user data and interpolating it into prompts, which presents an indirect injection risk. \n
- Ingestion points: In Step 1, the agent is directed to transcribe every legible string from user-supplied images verbatim. \n
- Boundary markers: The skill uses structural headers like 'TEXT:' and 'CAMERA:' in Step 3 to delimit inputs, which provides minimal isolation against malicious text content. \n
- Capability inventory: The skill has access to several media generation tools (Seedance 2.5, GPT-Image-2, Nano-Banana-2) and performs network operations (HTTP PUT) to upload image data to pre-signed URLs. \n
- Sanitization: There are no instructions provided to sanitize, validate, or filter the transcribed text before it is quoted in the prompt structure.
Audit Metadata