narrated-multi-scene
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the agent to install multiple dependencies from the author's official suite using the
npx skills addcommand. These tools provide specialized capabilities for image generation, video editing, and text-to-speech conversion. - [COMMAND_EXECUTION]: The workflow involves executing
ffmpegfor video concatenation and audio mixing, as well asffprobeto validate the duration of generated audio files. These commands are standard for media production tasks. - [INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided text for scene prompts and narration lines. To mitigate risks associated with untrusted content, the workflow incorporates mandatory human-in-the-loop review phases, requiring explicit user approval for the plan, stills, and final clips before proceeding to costly generation steps.
- Ingestion points: User-supplied scene descriptions, narration scripts, and style prompts in the intake phase.
- Boundary markers: The skill uses explicit phase gates like "approve plan" and "approve stills" to ensure the user reviews inputs and intermediate outputs.
- Capability inventory: File system access via
ffmpeg/ffprobeand network operations through integrated generation tools. - Sanitization: The process relies on manual user verification at multiple stages of the workflow.
Audit Metadata