remotion-video-toolkit
Pass
Audited by Gen Agent Trust Hub on Sep 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill describes architecture for dynamic video generation that processes untrusted external data.
- Ingestion points:
rules/calculate-metadata.md(fetches from user-provided URLs),rules/rendering.md(Express server taking user props via POST),rules/lottie.md(fetches Lottie JSON from external URLs). - Boundary markers: Absent; no instructions provided to delimit or ignore instructions within external data.
- Capability inventory:
renderMedia(rules/rendering.md) executes code to generate frames;extractFrames(rules/extract-frames.md) andcanDecode(rules/can-decode.md) process media from remote URLs. - Sanitization: No sanitization or validation of external content is shown in the examples.
- [EXTERNAL_DOWNLOADS]: The skill references an unknown third-party library 'mediabunny' for media processing.
- Evidence: Multiple files (e.g.,
rules/can-decode.md,rules/extract-frames.md,rules/get-video-duration.md) contain code snippets importing from the 'mediabunny' package, which is not an official Remotion library or a well-known public package. - [COMMAND_EXECUTION]: Instructions guide the agent through executing shell commands for project management and cloud deployment.
- Evidence:
rules/rendering.mdshowsnpx remotion render ...and commands for AWS Lambda and Google Cloud Run deployment (npx remotion lambda ...,npx remotion cloudrun ...).
Audit Metadata