The Agent Skills Directory

[SAFE]: The skill consists of structured markdown instructions and reference materials that guide the agent through a rigorous debugging workflow. It does not contain executable malicious instructions or hidden commands.- [SAFE]: The included utility script (evals/scripts/grade.py) is a benign Python tool for evaluating response quality. It uses standard libraries to perform text-based validation and does not engage in network operations or dynamic code execution.- [SAFE]: The skill explicitly instructs the agent to consult a human when changes impact security behavior or public contracts, reinforcing a safe operational model.- [SAFE]: No obfuscation techniques, base64-encoded payloads, or unauthorized data access patterns were identified across any of the provided files.- [SAFE]: The workflow emphasizes validating data at system boundaries and maintaining invariants, which are foundational principles of secure software development.

bug-fixing