previs-rerender
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes shell commands to locate its core dependencies and executes specific binaries (
curl,jq) to validate inputs and construct JSON payloads. This is part of its documented primary functionality. - [EXTERNAL_DOWNLOADS]: Employs
curlto perform a ranged read check on user-provided video URLs. This verifies the accessibility of remote content without downloading or storing the data locally, serving as a pre-flight check for the video processing job. - [INDIRECT_PROMPT_INJECTION]: The skill acts as an ingestion point for external video URLs and user-supplied prompts. It mitigates injection risks by using structured prompt templates that explicitly define boundaries for the AI, instructing it to prioritize the reference video's motion and layout cues over external influence.
Audit Metadata