video-editor-setup
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The instructions include explicit security constraints regarding sensitive data management. It strictly prohibits the agent from printing, echoing, or reading the
system/keys.envfile into the conversation, which is a key best practice for secret handling. - [SAFE]: The skill follows a collaborative 'human-in-the-loop' model, instructing the agent to build scripts step-by-step with the user and to ask for permission before any software installations.
- [SAFE]: The proposed folder structure and 'Hard Rules' prioritize data integrity, specifically forbidding the modification of raw source footage and limiting file operations to a dedicated workspace.
- [SAFE]: No evidence of prompt injection, obfuscation, or unauthorized network activity was detected. The external tools mentioned (ffmpeg, DaVinci Resolve, ElevenLabs) are standard in the specified industry context.
Audit Metadata