skills/simota/agent-skills/cue/Gen Agent Trust Hub

cue

Pass

Audited by Gen Agent Trust Hub on Oct 22, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection as it processes external data from user briefs, product specifications from other agents, and live product UI via Playwright browser control. It has extensive capabilities including browser navigation, JavaScript injection, and file system writes.
  • Ingestion points: Navigates to arbitrary URLs for recording; ingests narrative materials and specifications from agents such as 'Saga' and 'Scribe'.
  • Boundary markers: The skill instructs the agent to define audience and goals first, but does not provide explicit prompt delimiters to isolate content during browser execution.
  • Capability inventory: Extensive browser control via Playwright (page.screencast, page.goto, page.evaluate), file writing to project directories, and documented examples of shell execution for media processing.
  • Sanitization: No specific sanitization or filtering logic is documented for handling external inputs during the recording workflow.
  • [EXTERNAL_DOWNLOADS]: The skill provides instructions and patterns for installing necessary third-party tools, specifically the 'ffmpeg-quality-metrics' package via pip and the 'ffmpeg' utility via system package managers (e.g., apt-get). These are well-known, industry-standard tools required for the skill's primary function of video quality analysis and production.
  • [COMMAND_EXECUTION]: The reference documentation includes implementation patterns that utilize shell command execution for video transcoding and loudness normalization using FFmpeg. It also provides examples of automated environment setup in CI/CD contexts. These instructions are presented as standard operational procedures for the domain of video engineering.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 22, 2026, 01:06 PM