creativeclaw-seedance-2-5
Pass
Audited by Gen Agent Trust Hub on Sep 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill provides legitimate instructions for video generation and media management using the Creative Claw platform. All identified external resources and tool endpoints are associated with the vendor's own infrastructure or well-known cloud services.
- [INDIRECT_PROMPT_INJECTION]: The skill provides a surface for processing untrusted external media (images, videos, audio) which is inherent to its primary purpose of cinematic video generation.
- Ingestion points: Media assets are ingested via specific URL fields (image_urls, video_urls, audio_urls) and specialized import tools (import_chatgpt_media).
- Boundary markers: The workflow uses specific reference tokens (@ImageN, @VideoN) and descriptive character/identity protection prompts to guide model behavior, though it does not explicitly define technical boundaries for data-embedded instructions.
- Capability inventory: The skill has access to media assembly tools (merge_media, assemble_film), film project management (update_film_project), and network asset management operations (upload_asset).
- Sanitization: The instructions do not specify automated sanitization or filtering of external media content or metadata.
Audit Metadata