youtube-thumbnail-design
Pass
Audited by Gen Agent Trust Hub on Aug 28, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill guides the user to install the
belt-sh/clipackage and references installation instructions on GitHub (raw.githubusercontent.com/inference-sh/skills). These resources are required for the AI image generation features described in the skill. - [COMMAND_EXECUTION]: The skill provides shell examples using the
beltcommand-line interface. Security is maintained through theallowed-toolsconfiguration, which limits the agent to only executing commands that begin withbelt, preventing arbitrary command execution. - [INDIRECT_PROMPT_INJECTION]: The skill creates prompts for image generation which could be a vector for indirect injection. 1. Ingestion points: Prompts are defined in SKILL.md and passed as arguments to the
belttool. 2. Boundary markers: The commands use a structured JSON format for the--inputargument to separate parameters from prompt text. 3. Capability inventory: The skill's active capabilities are restricted to thebelttool via Bash as specified in SKILL.md. 4. Sanitization: While no specific input filtering is shown, the restricted tool access limits the potential impact of any injected content.
Audit Metadata