ai-image-generation
Pass
Audited by Gen Agent Trust Hub on Nov 18, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references installation instructions from the inference-sh GitHub organization and suggests adding the belt-sh/cli skill. These resources are standard components of the platform the skill is designed to interface with.
- [COMMAND_EXECUTION]: The skill utilizes the belt CLI to perform operations such as authentication and running AI model applications. Access is explicitly restricted to the belt utility via the allowed-tools configuration in the frontmatter, ensuring the agent cannot execute arbitrary shell commands.
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied text prompts that are passed as arguments to image generation models.
- Ingestion points: User prompts are ingested and passed into the --input flag of the belt app run command in SKILL.md.
- Boundary markers: Examples demonstrate the use of JSON structures to encapsulate user input, providing a clear boundary between command arguments and data.
- Capability inventory: The skill's capabilities are limited to executing the belt command (SKILL.md).
- Sanitization: The skill relies on JSON structure for data passing; no additional sanitization or instruction-ignoring delimiters are explicitly defined in the provided templates.
Audit Metadata