action-continuity-choreography

Pass

Audited by Gen Agent Trust Hub on Nov 14, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill references external URLs to goodcase.ai, x.com, and mp.weixin.qq.com along with direct links to media files (.mp4, .jpg, .webm) to serve as visual anchors. The instructions explicitly direct the agent to use available browser, image, or video tools to inspect these remote resources before drafting content.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to process and adapt 'Prompt excerpts' stored in references/cases.md. These excerpts contain imperative instructions meant for video models. While the skill provides a structured 'Preserve / Replace / Avoid' framework to handle this data, it represents a surface where complex external prompt data is ingested into the agent's context. The current content is focused on video choreography and does not contain malicious overrides.
  • [SAFE]: The skill incorporates strong safety and attribution guidelines, including prohibitions against making external account changes without approval, mandates for preserving creator attribution, and instructions to avoid impersonating original creators.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 14, 2026, 06:54 PM