landing-page-design
Warn
Audited by Gen Agent Trust Hub on Aug 12, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references remote installation instructions hosted at github.com/inference-sh/skills.
- [REMOTE_CODE_EXECUTION]: The skill prompts the installation of the belt-sh/cli package, which introduces external code from an unverified source into the runtime environment.
- [COMMAND_EXECUTION]: The skill utilizes the belt CLI to perform image generation and searches, involving shell command execution with user-defined parameters.
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection via its ingestion of untrusted data in command arguments. Ingestion points: --input fields in belt commands. Boundary markers: Single quotes. Capability inventory: Shell execution (belt tool). Sanitization: None identified.
Audit Metadata