remotion

Pass

Audited by Gen Agent Trust Hub on Sep 16, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill downloads screenshots from Stitch projects via Google Cloud Storage URLs (storage.googleapis.com) using curl or web_fetch as part of its asset retrieval process.\n- [COMMAND_EXECUTION]: The skill executes shell commands to set up the Remotion environment, install dependencies (npm install), and render the final video using the Remotion CLI.\n- [DYNAMIC_EXECUTION]: The skill dynamically generates TypeScript/React components and a JSON manifest based on design metadata to construct the video composition at runtime.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from Stitch project titles and screen descriptions to create video text overlays, which could theoretically contain malicious instructions targeted at the agent processing the data.\n
  • Ingestion points: Data retrieved from list_projects, list_screens, and get_screen tool outputs in SKILL.md.\n
  • Boundary markers: No specific delimiters or "ignore instructions" warnings are utilized during the metadata extraction phase.\n
  • Capability inventory: File system write, shell command execution, and network access capabilities are present in the skill's defined workflow.\n
  • Sanitization: The skill relies on standard React text node rendering for the overlays defined in the ScreenSlide.tsx template.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 16, 2026, 07:42 AM