gpt-image-sprite-sheet
Pass
Audited by Gen Agent Trust Hub on Nov 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses a local Bash script
scripts/slice.shto perform image processing tasks such as cropping, background keying, and generating GIF/atlas files. This script relies onImageMagick(specifically themagickbinary) and standard shell utilities. - [EXTERNAL_DOWNLOADS]: The skill interacts with the vendor's API (
api.varg.ai) for image generation and references assets hosted on the vendor's storage (s3.varg.ai). These are authorized vendor resources. - [INDIRECT_PROMPT_INJECTION]: The skill takes user-supplied action descriptions and character images as input, interpolating them into a prompt for the image generation model.
- Ingestion points: User-provided action text and character image files entered via the
promptandfilesfields in theimagetool call. - Boundary markers: The skill template uses Japanese and English bracketed indicators but lacks strong multi-token delimiters for user-interpolated content.
- Capability inventory: The skill possesses
Read,Write,Edit, andBashexecution capabilities for processing files and generating output data. - Sanitization: No specific input sanitization or validation logic is applied to the user-supplied prompt segments before model submission.
Audit Metadata