recording
Pass
Audited by Gen Agent Trust Hub on Jul 13, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: No prompt injection or behavior override patterns were detected. The skill describes functional implementation details and correctly notes the boundaries of the agent's influence on system state.\n- [DATA_EXFILTRATION]: No unauthorized data access or exfiltration patterns were found. Media uploads are directed to the platform's internal API endpoints as part of core functionality.\n- [REMOTE_CODE_EXECUTION]: No remote code execution or dynamic code evaluation patterns were identified. The use of ffmpeg.wasm for client-side processing is standard practice for this use case.\n- [CREDENTIALS_UNSAFE]: No hardcoded credentials or sensitive configuration exposure was found.\n- [EXTERNAL_DOWNLOADS]: External integrations are limited to well-known services like Loom and the Chrome Web Store for intended features, such as importing recordings or offering extensions.
Audit Metadata