The Agent Skills Directory

[SAFE]: The skill defines a rigorous process for prompt engineering, including baseline preparation, scenario design, and bias-free execution using sub-agents.
[SAFE]: It includes environmental checks and skip-logic to handle platforms where sub-agent dispatch (the Agent tool) is disabled or restricted, ensuring it fails gracefully without compromising its core bias-free requirement.
[SAFE]: The instructions emphasize using the Agent tool for its intended purpose (simulating a 'blank-slate' user) and explicitly forbid manual self-rereading to prevent developer bias.
[SAFE]: No malicious patterns, obfuscation, data exfiltration, or credential harvesting were detected. The skill uses local execution and standard agent tools like Read and Agent to perform its tasks.

prompt-tuning