remotion
Pass
Audited by Gen Agent Trust Hub on Nov 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides documentation and examples for ingesting external data via the
calculateMetadatafunction andgetInputPropsAPI. This creates a surface where an attacker could provide data containing malicious natural language instructions designed to manipulate the AI agent's behavior during video processing. - Ingestion points: Example code in
SKILL.mdandreference.mdshowscalculateMetadatafetching data from/api/dataand the use ofgetInputProps()to retrieve CLI-supplied arguments. - Boundary markers: The provided code snippets do not include boundary markers or explicit instructions for the model to ignore potential injection content within the fetched data.
- Capability inventory: The skill enables significant file system influence (writing video frames, referencing static assets) and network capabilities (fetching data, rendering on AWS Lambda).
- Sanitization: No sanitization or validation logic is present in the examples to filter or escape the content retrieved from external sources.
Audit Metadata