The Agent Skills Directory

[SAFE]: The skill defines an orchestration process centered on the scientific method and human oversight. It includes phases for framing, design verification, and audit logging to ensure transparency.
[SAFE]: No malicious patterns, network exfiltration, or unauthorized credential access were detected. The skill uses environment-specific logic to toggle delegation features safely, only activating them when specific conditions are met.
[SAFE]: Explicit safety constraints are integrated into the instructions, such as preventing agents from granting their own human approval and requiring the preservation of user-modified files.

custom-workflow