The Agent Skills Directory

[SAFE]: The skill provides developer documentation and code templates for the ageval_sdk. It defines how to structure run.py and evaluator.py files for agent benchmarking.
[SAFE]: Includes an antipatterns.md file that explicitly warns against security risks such as reading local authentication files (e.g., ~/.codex/auth.json) or printing secrets, promoting safe coding practices.
[SAFE]: The provided code snippets demonstrate standard use of an SDK for task execution and reporting without any malicious command execution, network exfiltration, or prompt injection patterns.

ageval-sdk-harness