background-removal
Pass
Audited by Gen Agent Trust Hub on Aug 7, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Fetches installation instructions and supplementary skills from the official inference-sh GitHub repository and skill registry.
- [COMMAND_EXECUTION]: Executes shell commands using the belt CLI to perform image inference tasks and manage user authentication via belt login.
- [PROMPT_INJECTION]: The skill processes user-provided image URLs that are interpolated into JSON payloads for shell commands, creating a surface for indirect prompt injection. 1. Ingestion points: image_url parameter in belt app run commands within SKILL.md. 2. Boundary markers: No delimiters or ignore-instructions warnings are used to wrap the user-supplied URL. 3. Capability inventory: Includes shell command execution, network requests, and file writing. 4. Sanitization: No evidence of validation or sanitization for the external URL input.
Audit Metadata