ai-video-generation
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill facilitates the installation of external components and references remote configuration files from the inference-sh ecosystem.
- Instructs the user to install the belt CLI skill using
npx skills add belt-sh/cli. - References several sub-skills (e.g.,
inference-sh/skills@google-veo,inference-sh/skills@seedance) for specific model capabilities. - Links to remote installation instructions at
https://raw.githubusercontent.com/inference-sh/skills/refs/heads/main/cli-install.md. - [INDIRECT_PROMPT_INJECTION]: The skill exposes a surface for indirect prompt injection due to its core functionality of processing external data and user-defined prompts.
- Ingestion points: The
belt app runcommand examples inSKILL.mdingest externalimage_url,audio_url, andvideo_urlvalues, alongsidepromptstrings. - Boundary markers: There are no explicit delimiters or specific instructions to the model to ignore potential commands embedded within the processed media files or URLs.
- Capability inventory: The skill uses the
Bashtool restricted tobeltcommands, which perform network operations to interact with remote inference APIs. - Sanitization: No sanitization or validation of the provided URLs or prompt contents is described within the skill instructions.
Audit Metadata