framevideo-media

Pass

Audited by Gen Agent Trust Hub on Jul 21, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute various commands using the npx framevideo CLI tool. This is a vendor-provided tool (chanjing-ai) used for text-to-speech, transcription, and asset management.
  • [EXTERNAL_DOWNLOADS]: The skill downloads model weights for Kokoro TTS, Whisper, and u2net background removal during its initial run. These are sourced from well-known repositories and organizations including OpenAI.
  • [EXTERNAL_DOWNLOADS]: The skill fetches audio assets (BGM/SFX) from the Chanjing platform (chanjing.cc), which is the vendor's official service.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user data, such as scripts for TTS or audio/video files for transcription. While this creates a potential surface for indirect prompt injection, it is managed through standard CLI tools and serves the skill's primary purpose.
  • [CREDENTIALS_UNSAFE]: The skill includes instructions for authentication using npx framevideo auth login. This is used to manage access to the vendor's music library and is standard practice for platform integration.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 21, 2026, 06:24 AM
Security Audit — agent-trust-hub — framevideo-media