imagegen
Pass
Audited by Gen Agent Trust Hub on Sep 10, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill utilizes well-known Python packages
openaiandpillowfor API interaction and image manipulation. These are standard dependencies fetched from the official Python package registry. - [COMMAND_EXECUTION]: The skill executes a bundled Python script (
scripts/image_gen.py) to interface with the OpenAI API. This is the primary intended function of the skill for performing deterministic image generation and batch operations. - [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied prompts and images to generate new visual content. While it employs a structured augmentation template, it does not include specific delimiters to protect the image model from instructions embedded within the user's input data.
- Ingestion points: User-provided prompts and image/mask paths processed in
SKILL.mdandscripts/image_gen.pyviaargparseand JSONL inputs. - Boundary markers: Absent; user prompts are interpolated into a structured text spec without specific isolation markers.
- Capability inventory: Network access to OpenAI API endpoints and local file system access for reading/writing images and temporary JSONL files.
- Sanitization: None; user inputs are directly incorporated into the final prompt augmentation spec.
Audit Metadata