The Agent Skills Directory

[SAFE]: The skill implements a robust framework for agent prompt design, emphasizing verified authority and measurable outcomes while explicitly prohibiting the agent from assuming edit authority based on ambiguous user inputs.- [SAFE]: The included validation utility (scripts/check.py) performs static analysis of the repository to enforce contract compliance and uses regular expressions to prevent the accidental inclusion of hardcoded host system paths (e.g., /Users, /home), enhancing environment isolation.- [SAFE]: External resources and documentation snapshots are sourced from trusted official domains (developers.openai.com) and are accompanied by SHA-256 integrity hashes to ensure their authenticity and prevent tampering.- [SAFE]: The skill manages the risk of processing untrusted data by instructing the agent to strictly distinguish between user requirements, observed repository facts, and verified external data, providing clear boundary markers for reasoning.- [SAFE]: No evidence of credential exposure, network exfiltration, or persistence mechanisms was found; all included scripts are limited to local, read-only validation tasks using standard libraries.

prompt-engineering