magic-hour-talking-video

Pass

Audited by Gen Agent Trust Hub on Sep 12, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied scripts for voice generation, creating an opportunity for malicious instructions to be embedded in the content.\n
  • Ingestion points: Ingests user-provided text scripts and media file paths (SKILL.md).\n
  • Boundary markers: The skill lacks delimiters or instructions to ignore embedded prompts; instructions specifically require the agent to preserve the "exact script."\n
  • Capability inventory: Performs network operations via the Magic Hour API (job creation, status retrieval) and reads local media files for upload (SKILL.md).\n
  • Sanitization: The instructions do not specify any validation or sanitization for the user-provided text content.\n- [EXTERNAL_DOWNLOADS]: The skill references setup documentation and API schemas from the vendor's official GitHub repository and documentation site.\n
  • Evidence: References to https://github.com/magichourhq/magic-hour-mcp/blob/main/user.md and https://docs.magichour.ai/api-reference/openapi.json for configuration and reference.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 12, 2026, 05:09 PM
Security Audit — agent-trust-hub — magic-hour-talking-video