paw-cra-video-longform

Pass

Audited by Gen Agent Trust Hub on Sep 18, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill fetches stock footage from Pexels, synthesized audio from ElevenLabs, and generated video clips from Fal.ai. These are well-known services integrated into the production workflow to support asset acquisition.
  • [COMMAND_EXECUTION]: The workflow orchestrates local commands including ffmpeg for assembly and egaki CLI for AI video generation. The validate-video.py script uses subprocess.run to call ffprobe for metadata verification and ffmpeg for loudness measurement.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied briefs and script data to generate prompts for AI models, creating a vulnerability surface for indirect injection.
  • Ingestion points: SKILL.md Stage 1 (Brief Intake) and Stage 2 (Script Development).
  • Boundary markers: No specific delimiters or safety instructions are defined for the interpolated script content in the prompt templates.
  • Capability inventory: The skill has access to the filesystem via ffmpeg and the network via curl and specialized API clients.
  • Sanitization: Input text is not sanitized before being interpolated into AI generation prompts.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 18, 2026, 01:29 PM