remotion
Warn
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- Prompt Injection (MEDIUM): The file references/tailwind.md instructs the agent to fetch documentation from an external URL (https://www.remotion.dev/docs/tailwind) specifically to obtain instructions, which creates a vector for an external source to influence agent behavior.
- Indirect Prompt Injection (MEDIUM): The skill documents multiple ingestion surfaces for untrusted data without providing sanitization or boundary markers. Evidence: Ingestion points include references/lottie.md (remote JSON), references/calculate-metadata.md (remote API props), and references/import-srt-captions.md (remote SRT fetching). Capability inventory: The skill enables complex video rendering and logic based on these external inputs. Sanitization: None provided in examples.
- External Downloads (MEDIUM): The skill references and suggests installation of various packages (mediabunny, @remotion/*) from sources not included in the Trusted External Sources list. This includes patterns that fetch and process remote media directly via UrlSource in references/extract-frames.md and references/get-video-dimensions.md.
Audit Metadata