faceless-explainer
Pass
Audited by Gen Agent Trust Hub on Sep 26, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill orchestrates the production process by running a series of local Node.js utility scripts such as build-frame.mjs and audio.mjs, and using the npx tool to invoke the hyperframes CLI for project initialization and rendering.
- [EXTERNAL_DOWNLOADS]: During execution, the skill fetches the hyperframes package and references the GSAP animation library from the JSDelivr CDN in its generated HTML files. These are well-known resources from trusted CDNs.
- [DYNAMIC_EXECUTION]: The audio.mjs script uses an environment variable HF_MEDIA_ENGINE to resolve the location of a media processing engine at runtime, which is then executed via spawnSync. This is a standard adapter pattern for narration services.
- [INDIRECT_PROMPT_INJECTION]: 1. Ingestion points: Raw user-provided text is saved to capture/extracted/visible-text.txt during Step 1. 2. Boundary markers: The skill does not employ specific delimiters or instructions to ignore embedded prompts within the source text. 3. Capability inventory: The skill can write files, execute scripts, and render media files. 4. Sanitization: There is no explicit sanitization or filtering of the user text before it is processed by the agent to create the storyboard.
Audit Metadata