action-continuity-choreography-by-1lcb8j2
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external sources and user-provided media, creating a surface for indirect prompt injection.
- Ingestion points: The agent reads prompt excerpts and accesses media URLs (images/videos) from
references/cases.md. It also requests target media from the user, including opening/ending frames, reference images, source videos, and motion guides. - Boundary markers: The skill utilizes a structured 'reference contract' (Preserve / Replace / Avoid) and explicit instructions to 'Distinguish observed evidence from your own recommendation', which serve as logical boundaries to separate reference data from the final output.
- Capability inventory: The skill instructs the agent to use image, video, or browser tools to inspect external media and then generate text-based artifacts (video prompts, motion briefs).
- Sanitization: There are no technical sanitization steps for the ingested media content or the external prompt excerpts beyond the agent's internal safety filters.
Audit Metadata