pixelkiln
Pass
Audited by Gen Agent Trust Hub on Oct 31, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill correctly manages credentials by instructing the agent to never print, inspect, or commit API keys (
PIXELLAB_API_KEY,RD_API_KEY,SCENARIO_SDK_API_KEY, etc.) to version control, recommending the use of.env.localand environment variables instead. - [SAFE]: All major generative operations require explicit human review and approval (via the
pixelkiln refine approvecommand), ensuring the agent does not autonomously commit to paid work or finalize unverified output. - [SAFE]: The skill manages remote configurations (recipes) through a verification-first workflow, instructing users and agents to inspect recipes before installation and installation via pinned versions.
- [SAFE]: Security guidance is provided regarding local service connectivity, specifically advising that unauthenticated ComfyUI servers should remain isolated from public networks.
Audit Metadata