ai-image-gen

Pass

Audited by Gen Agent Trust Hub on Apr 15, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONCREDENTIALS_UNSAFEPROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill provides legitimate instructions for an image processing workflow and does not contain any malicious patterns or unauthorized access attempts.
  • [EXTERNAL_DOWNLOADS]: Fetches the 'ai-img' tool from the author's GitHub repository and suggests global installation via Bun. These are standard deployment methods for the developer's software and are documented neutrally.
  • [COMMAND_EXECUTION]: Defines several commands to execute the 'ai-img' CLI locally for processing images and prompts. These operations are limited to the tool's intended purpose and scope.
  • [CREDENTIALS_UNSAFE]: References the use of environment variables like OPENAI_API_KEY and FAL_API_KEY. The instructions correctly advise users to export these keys, which is a security best practice for managing sensitive credentials.
  • [PROMPT_INJECTION]: The skill processes untrusted user prompts and JSONL data files, which creates an indirect prompt injection surface. This is a characteristic of the image generation use case and is handled by the downstream AI provider's safety filters.
  • Ingestion points: 'prompt' arguments in CLI commands and the 'jobs.jsonl' input file.
  • Boundary markers: None.
  • Capability inventory: Execution of CLI commands that perform network requests to AI image generation APIs.
  • Sanitization: Relies on the internal logic of the 'ai-img' tool and the safety protocols of the integrated AI providers.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 15, 2026, 01:59 AM