video-editing
Pass
Audited by Gen Agent Trust Hub on Apr 1, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: Executes FFmpeg, Node.js, and shell scripts via the
Bashtool to perform video analysis, cutting, and rendering tasks. - [EXTERNAL_DOWNLOADS]: Fetches and executes Node.js packages using
npmandnpxto facilitate video composition through the Remotion framework. - [DATA_EXFILTRATION]: Performs authenticated network requests to the ElevenLabs API using a key retrieved from the environment for voiceover generation.
- [PROMPT_INJECTION]: Potential surface for indirect prompt injection as the skill processes
transcript.txt(derived from user video) to guide the automated editing process. - Ingestion points: Reads
transcript.txtinSKILL.md(Phase 2). - Boundary markers: None identified; the skill relies on the AI's general judgment of the transcribed text.
- Capability inventory:
Bash(file and process management) andWrite(file system access) are available throughout the pipeline. - Sanitization: None; content is processed directly to generate a structured Edit Decision List (EDL).
Audit Metadata