create-video

Pass

Audited by Gen Agent Trust Hub on Sep 13, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied text prompts to generate videos via the HeyGen API, creating a surface for potential indirect instructions to be passed to the underlying video generation model.\n
  • Ingestion points: User-provided prompts are used as input for the Video Agent API in SKILL.md and references/video-agent.md.\n
  • Boundary markers: The documentation and code snippets do not include explicit boundary markers or instructions for the agent to disregard commands embedded within the user prompt.\n
  • Capability inventory: The skill documents capabilities for reading local files (fs.readFileSync in references/assets.md), writing files (fs.writeFileSync in references/video-status.md), and performing network operations (curl, fetch, requests in multiple reference files) to interact with the HeyGen API.\n
  • Sanitization: No explicit sanitization or validation of the user prompt is present in the provided reference snippets.\n- [EXTERNAL_DOWNLOADS]: The skill documentation describes interactions with official HeyGen service endpoints, including api.heygen.com, upload.heygen.com, and files.heygen.ai for video generation and asset management.\n- [DATA_EXFILTRATION]: The skill includes instructions and code samples for reading local files and uploading them to HeyGen's infrastructure, as well as downloading generated video content. These operations are intended functionality for the service's asset management workflow.\n- [COMMAND_EXECUTION]: The skill provides reference curl commands intended for use by the agent or developer to interact with the HeyGen API endpoints.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 13, 2026, 12:09 AM
Security Audit — agent-trust-hub — create-video