make-a-video
Pass
Audited by Gen Agent Trust Hub on Apr 19, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes shell commands using
ffmpegto extract specific video frames for quality assurance andnpxto interact with the HyperFrames CLI tools for project initialization, linting, and rendering. - [EXTERNAL_DOWNLOADS]: The skill invokes
npx, which is designed to download and execute packages from the npm registry if they are not already cached locally. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it transforms untrusted user input (scripts and outlines) into executable code (HTML and JavaScript) for video compositions.
- Ingestion points: User input captured during the interview gates (Gates 1, 2, and 3) in SKILL.md.
- Boundary markers: The workflow enforces a manual review gate where the agent must present a BRIEF.md and STORYBOARD.md to the user for explicit approval before building.
- Capability inventory: File system writes for scaffolding and composition creation, shell command execution via ffmpeg and npx, and local network serving (all occurring via instructions in SKILL.md).
- Sanitization: The skill lacks explicit logic to sanitize or escape user-provided text before it is interpolated into the generated composition files.
Audit Metadata