dreamina-video-multi-frame
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- General Analysis (SAFE): The code implements a standard API client for the Jianying video generation service. It uses the
requestslibrary to send structured JSON data to a legitimate domain (jimeng.jianying.com).\n- Data Handling (SAFE): While the skill handles session IDs, they are used solely for authentication with the target API as required for the functionality. No evidence of unauthorized data exposure or exfiltration was found.\n- Indirect Prompt Injection (LOW): The skill accepts user-provided prompts and image URIs which are passed directly to an external API.\n - Ingestion points:
image_uri_listandprompt_listinSKILL.md\n - Boundary markers: Absent\n
- Capability inventory:
requests.postnetwork call\n - Sanitization: Absent\n
- Summary: The risk is limited to poisoning the input of the external video generation tool.
Audit Metadata