ai-image-generation
Pass
Audited by Gen Agent Trust Hub on Jul 23, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the
beltCLI tool to run image generation and editing tasks. Execution is scoped to thebeltcommand through theallowed-toolsconfiguration, minimizing risk from arbitrary shell execution. - [EXTERNAL_DOWNLOADS]: Instructions include installing the
beltCLI vianpx skills add belt-sh/cliand referencing related skills from theinference-shorganization. These represent standard setup procedures for the tool's ecosystem. - [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it interpolates user-provided text into the
--inputargument of shell commands. - Ingestion points: User-provided prompts and image URLs are passed to
belt app runinSKILL.mdexamples. - Boundary markers: Absent; prompts are directly embedded in JSON strings.
- Capability inventory: Uses
Bashrestricted to thebeltcommand as defined in the frontmatter. - Sanitization: No explicit sanitization or validation of the input prompt is performed by the skill itself, relying on downstream model/service filters.
Audit Metadata