跳转至

Eval 命令

管理基于评估的开发工作流。

用法

/eval [define|check|report|list|clean] [feature-name]

评估定义

/eval define feature-name

创建新的评估定义:

  1. 在 .claude/evals/feature-name.md 中创建模板:
## EVAL: feature-name
Created: $(date)

### Capability Evals
- [ ] [기능 1에 대한 설명]
- [ ] [기능 2에 대한 설명]

### Regression Evals
- [ ] [기존 동작 1이 여전히 작동함]
- [ ] [기존 동작 2이 여전히 작동함]

### Success Criteria
- capability eval에 대해 pass@3 > 90%
- regression eval에 대해 pass^3 = 100%
  1. 提示用户输入具体标准

评估检查

/eval check feature-name

对功能运行评估:

  1. 从 .claude/evals/feature-name.md 中读取评估定义
  2. 对于每个能力评估(capability eval):
  3. 尝试验证标准
  4. 记录 PASS/FAIL
  5. 将尝试记录到 .claude/evals/feature-name.log
  6. 对于每个回归评估(regression eval):
  7. 运行相关测试
  8. 与基线比较
  9. 记录 PASS/FAIL
  10. 报告当前状态:
EVAL CHECK: feature-name
========================
Capability: X/Y passing
Regression: X/Y passing
Status: IN PROGRESS / READY

评估报告

/eval report feature-name

生成综合评估报告:

EVAL REPORT: feature-name
=========================
Generated: $(date)

CAPABILITY EVALS
----------------
[eval-1]: PASS (pass@1)
[eval-2]: PASS (pass@2) - 재시도 필요했음
[eval-3]: FAIL - 비고 참조

REGRESSION EVALS
----------------
[test-1]: PASS
[test-2]: PASS
[test-3]: PASS

METRICS
-------
Capability pass@1: 67%
Capability pass@3: 100%
Regression pass^3: 100%

NOTES
-----
[이슈, 엣지 케이스 또는 관찰 사항]

RECOMMENDATION
--------------
[SHIP / NEEDS WORK / BLOCKED]

评估列表

/eval list

显示所有评估定义:

EVAL DEFINITIONS
================
feature-auth      [3/5 passing] IN PROGRESS
feature-search    [5/5 passing] READY
feature-export    [0/4 passing] NOT STARTED

参数

$ARGUMENTS: - define <name> - 创建新的评估定义 - check <name> - 运行并检查评估 - report <name> - 生成完整报告 - list - 显示所有评估 - clean - 清理旧评估日志(保留最近 10 次执行)

本页原文 Markdown:在 AtomGit 查看·内容源自开源项目 ev/ECC