The Agent Skills Directory

[SAFE]: The skill serves as a process guide for performance optimization. It enforces best practices such as measuring baselines before changes, defining numeric targets, and ensuring identical measurement protocols to prevent false performance gains.
[SAFE]: The instructions mandate running full test suites after every individual change to prevent correctness regressions, which aligns with secure and reliable software development practices.
[SAFE]: It utilizes a structured template (references/task-template.md) to guide the agent through recording hypotheses, change ledgers, and evidence, promoting transparency and auditability of the agent's actions.
[SAFE]: The skill references an external AGENTS.md file for resolving benchmark and test commands, which is a standard platform mechanism for providing a controlled execution environment.

write-performance