Transforms vague performance requests into measurable optimization loops with baselines, variants, and correctness gates.
How this skill is triggered — by the user, by Claude, or both
Slash command
/everything-claude-code:benchmark-optimization-loopThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
使用此技能将"使其快 20 倍"或"尝试 50 次递归优化"转换为可以实际改进系统的有界测量循环。
使用此技能将"使其快 20 倍"或"尝试 50 次递归优化"转换为可以实际改进系统的有界测量循环。
在以下条件存在之前不要优化:
如果用户要求不切实际的目标,保持雄心但使循环有界且可测量。
像这样跟踪变体:
变体 | 假设 | 命令 | 时间 | 正确? | 备注
baseline | 当前路径 | npm run job | 120s | 是 | 稳定
batch-500 | 更少的往返 | npm run job -- --batch 500 | 42s | 是 | 获胜者
parallel-8 | 更多工作线程 | npm run job -- --workers 8 | 31s | 否 | 速率限制
对于递归或超参数工作:
使用"最佳测量安全变体"之类的短语,而非"全局最优",除非搜索空间实际上是详尽的。
变体在以下情况之前不能成为新默认值:
npx claudepluginhub aaione/everything-claude-code-zhConverts vague performance requests into bounded, measured optimization loops with baselines, variant tables, correctness gates, and rollback-safe promotion.
Runs autonomous optimization loops with 3 parallel agents per round, using shell metric commands, git worktrees, guards, and baselines to iteratively improve code performance.
Guides interactive setup of optimization goals, metrics, and scope; runs autonomous git-committed experiment loops: code changes, testing, measurement, keep improvements or revert. For performance tuning in git repos.