charles-reel-plans
Pass
Audited by Gen Agent Trust Hub on Nov 19, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The kit utilizes
subprocess.runinkit/build.pyto execute system utilities such asffmpegandffprobe. These tools are used for legitimate media processing tasks, including normalizing audio and determining video durations. - [EXTERNAL_DOWNLOADS]: The skill fetches assets, metadata, and configuration from well-known technology services including Wikimedia Commons, Wikipedia, and the Internet Archive (Wayback Machine). These operations are essential for the skill's primary purpose of sourcing historical and reference media.
- [DYNAMIC_EXECUTION]: Headless Google Chrome or Chromium is used by
kit/capture.pyto capture screenshots of web references and generate high-resolution preview frames for the video plan. This execution is scoped to a temporary user profile to mitigate cross-session data leakage. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from external URLs (such as social media posts and web pages), which presents a surface for indirect prompt injection. However, the skill implements robust sanitization, including HTML escaping and secure JSON encoding, when generating its final output (the Artifact page).
Audit Metadata