video-ad-specs

Pass

Audited by Gen Agent Trust Hub on Sep 2, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill requires the installation of an external CLI tool ('belt-sh/cli') via npx and refers to installation instructions hosted on a GitHub repository (inference-sh/skills).
  • [COMMAND_EXECUTION]: The skill provides numerous shell command templates utilizing the 'belt' CLI to perform tasks such as user authentication, running AI model inference for video/audio generation, and media processing operations (merging and captioning).
  • [INDIRECT_PROMPT_INJECTION]: The skill includes an attack surface where user-supplied natural language prompts are interpolated into JSON payloads for shell commands.
  • Ingestion points: User-provided prompts in the 'prompt' field of JSON inputs for various 'belt app run' commands (detected in SKILL.md).
  • Boundary markers: None identified; user input is directly embedded into JSON strings.
  • Capability inventory: The skill has access to the 'Bash' tool with 'belt' command execution (specified in SKILL.md).
  • Sanitization: No explicit sanitization or validation of the user-provided prompt content is described.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 2, 2026, 02:24 PM
Security Audit — agent-trust-hub — video-ad-specs