magic-hour-talking-video
Pass
Audited by Gen Agent Trust Hub on Nov 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided content such as scripts, audio recordings, and video files, which are potential ingestion points for indirect instructions. The instructions include mitigations such as strictly preserving the provided script and excluding stage directions from the speech generation process.
- Ingestion points: User-supplied portraits, audio, video, and text scripts (SKILL.md, references/localization.md).
- Boundary markers: Directives to preserve the exact script and exclude non-spoken text (SKILL.md).
- Capability inventory: Tools for media upload, project status retrieval, and API calls to the video generation service (SKILL.md).
- Sanitization: No specific programmatic sanitization is described in the instructions.
- [EXTERNAL_DOWNLOADS]: The skill contains references to external documentation and example repositories on GitHub for developer setup and technical guidance. These are informational and do not involve runtime code execution.
- Evidence: Documentation links to magichourhq and vargHQ repositories in SKILL.md and references/localization.md.
Audit Metadata