happy-horse-video-edit
Pass
Audited by Gen Agent Trust Hub on Aug 12, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill requires the agent to execute shell commands to run the Python script
videogen.pyand install therequestslibrary. - [DATA_EXFILTRATION]: The skill uploads local media files to the AI Hive API at
https://ai-hive.iclip.cn/apiand stores a service API key in the local file~/.ai-hive/config.json. - [EXTERNAL_DOWNLOADS]: The script downloads resulting video and image files from URLs provided by the API to the local path
~/Downloads/AiHive. - [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied file paths for upload, creating a surface for potential misuse if an agent is tricked into uploading sensitive system files. Ingestion points: CLI arguments
--video,--image, and--audioinvideogen.py. Boundary markers: Absent. Capability inventory: File reading (open), network POST (requests), and browser interaction (webbrowser.open) inscripts/videogen.py. Sanitization: Absent.
Audit Metadata