opus-video-skills
Pass
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill directs users to install plugins and fetch templates from a third-party GitHub repository (
tuzhechen2005/opus-video-skills). While GitHub is a known service, the repository belongs to an unknown account outside of the trusted organization scope. - [INDIRECT_PROMPT_INJECTION]: The skill's workflow involves processing untrusted external data to generate video content and associated code, creating a surface for injection attacks.
- Ingestion points: User-provided project materials, music files, and lyrics (LRC) files used to determine timing and visual events.
- Boundary markers: The skill lacks explicit instructions or delimiters to isolate processed data from the agent's internal instructions.
- Capability inventory: The skill utilizes shell commands for FFmpeg encoding and Node.js rendering, and it writes executable JavaScript scene files to the local file system.
- Sanitization: No validation or sanitization routines are defined for the external inputs before they are interpolated into the code generation process.
- [DYNAMIC_EXECUTION]: The skill generates and executes JavaScript code at runtime (such as IIFE scenes and cue files) based on user descriptions. This programmatic generation of executable logic from natural language input represents a dynamic execution risk.
Audit Metadata