ai-image-generation
Pass
Audited by Gen Agent Trust Hub on Aug 31, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill is configured to use the belt command-line tool via Bash. The allowed-tools frontmatter explicitly limits the scope of shell execution to this specific CLI tool.
- [EXTERNAL_DOWNLOADS]: The skill provides links to external documentation and installation instructions hosted on the inference.sh website and the inference-sh GitHub organization. These resources are consistent with the integration's purpose.
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied prompt strings that are passed to the belt CLI as JSON input. While this represents an attack surface where malicious input could influence the behavior of remote image models, it is a necessary functional component of an AI image generation integration.
- Ingestion points: Command examples in SKILL.md ingest user data through the prompt field in the --input flag.
- Boundary markers: None are specified in the command examples.
- Capability inventory: Shell command execution via Bash(belt *).
- Sanitization: No explicit sanitization or filtering is described in the skill instructions.
Audit Metadata