prepare-video-publish

Warn

Audited by Gen Agent Trust Hub on Jul 19, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes multiple local Node.js and Python scripts to perform video processing and content orchestration. Additionally, it dynamically assembles JavaScript fragments from local files and executes them within a browser environment using the Ego Lite utility to automate publishing tasks.
  • [EXTERNAL_DOWNLOADS]: Transcription functionality is provided through the 'faster-whisper' library, which may download pre-trained AI model weights from trusted public repositories during initial setup.
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface when processing video transcripts. 1. Ingestion points: Transcripts are read from local files (transcriptPath) and used for generating title and description content. 2. Boundary markers: The agent is instructed to ground claims in the transcript but lacks explicit delimiters or warnings to ignore embedded instructions. 3. Capability inventory: The skill can execute shell commands (scripts/video-publish.mjs), perform file operations, and control browser sessions via ego-browser. 4. Sanitization: No explicit validation or filtering of transcript content is observed before it is used to influence the agent's output.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jul 19, 2026, 01:44 AM
Security Audit — agent-trust-hub — prepare-video-publish