remotion-video-toolkit

Pass

Audited by Gen Agent Trust Hub on Oct 30, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: Example 1 includes a bash script that iterates over user data in users.json and interpolates extracted values directly into a shell command (npx remotion render). This pattern is vulnerable to command injection if the input data contains shell metacharacters (e.g., backticks or dollar signs).- [INDIRECT_PROMPT_INJECTION]: The skill defines a surface where untrusted data from JSON files (users.json, props.json) is ingested and rendered into video components. It lacks guidance on sanitizing this data or using boundary markers to prevent the injection of malicious content into the rendered output.
  • Ingestion points: users.json in Example 1 and props.json in Task D.
  • Boundary markers: Not implemented in the provided templates.
  • Capability inventory: File system write access for video rendering and AWS Lambda deployment/execution capabilities.
  • Sanitization: No input validation or escaping logic is included in the documentation.- [EXTERNAL_DOWNLOADS]: The skill uses npx create-video@latest to bootstrap projects and installs various @remotion/* packages and other dependencies from the npm registry.- [DYNAMIC_EXECUTION]: The core functionality relies on rendering React components in a headless browser (Chrome), which involves dynamic JavaScript execution of code that may include user-provided props.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 30, 2026, 12:00 PM