The Agent Skills Directory

[SAFE]: Behavioral Safety Framework: The skill implements a robust self-verification process ('The Five Gates') designed to prevent the agent from taking destructive, irreversible, or unauthorized actions.
[SAFE]: Explicit Consent Requirements: It explicitly defines a list of high-risk actions—such as database deletions, force-pushes, and system-level changes—that require clear and specific user approval before execution.
[SAFE]: Anti-Hallucination Instructions: The skill directs the agent to verify facts, measurements, and system states using appropriate tools rather than guessing or inferring content from metadata.
[SAFE]: Error Handling and Verification: It provides guidance on interpreting full command outputs and verifying completion to avoid partial or incorrect task fulfillment.
[SAFE]: No Malicious Patterns: No indicators of prompt injection, data exfiltration, obfuscation, or unauthorized remote code execution were detected.

before-you-act