vs
Pass
Audited by Gen Agent Trust Hub on Nov 1, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill executes various local CLI tools to manage media assets. This includes running
ffmpegandffprobefor video assembly and metadata probing, as well as macOS-specific utilities likeqlmanageandsipsfor rasterizing title cards from SVG templates. - [EXTERNAL_DOWNLOADS]: The CLI performs network operations to interact with multiple AI service providers, including BytePlus (Seedance), Vercel AI Gateway, Google (Gemini/Lyria), and ElevenLabs. These connections are used to submit generation tasks and download resulting media clips, music, and narration.
- [CREDENTIALS_SAFE]: The documentation identifies several required sensitive API keys (e.g.,
ARK_API_KEY,GEMINI_API_KEY,ELEVENLABS_API_KEY) and correctly instructs the user to store them in a local.envfile. This follows established security best practices for secret management in development tools. - [INDIRECT_PROMPT_INJECTION]: The skill acts as a data-processing pipeline that ingests user-provided local files such as
shots.json,stills.json, andlines.tsv. While these files contain instructions passed to AI models, the processing is localized to the user's environment and the risk of unauthorized behavior is low, consistent with the tool's primary function as a content generation orchestrator.
Audit Metadata