---

name: eval-harness description: 用于 Claude Code 会话的正式评估框架，实现评估驱动开发 (EDD) 原则 origin: ECC tools: Read, Write, Edit, Bash, Grep, Glob

评估框架技能

用于 Claude Code 会话的正式评估框架，实现评估驱动开发 (EDD) 原则。

激活时机

• 为 AI 辅助工作流设置评估驱动开发 (EDD)
• 定义 Claude Code 任务完成的通过/失败标准
• 使用 pass@k 指标衡量代理可靠性
• 为提示或代理变更创建回归测试套件
• 跨模型版本基准测试代理性能

理念

评估驱动开发将评估视为"AI 开发的单元测试"：

• 在实现之前定义预期行为
• 在开发过程中持续运行评估
• 跟踪每次变更的回归
• 使用 pass@k 指标衡量可靠性

评估类型

能力评估

测试 Claude 是否能做到之前做不到的事情：

[CAPABILITY EVAL: feature-name]
Task: Description of what Claude should accomplish
Success Criteria:
  - [ ] Criterion 1
  - [ ] Criterion 2
  - [ ] Criterion 3
Expected Output: Description of expected result

回归评估

确保变更不会破坏现有功能：

[REGRESSION EVAL: feature-name]
Baseline: SHA or checkpoint name
Tests:
  - existing-test-1: PASS/FAIL
  - existing-test-2: PASS/FAIL
  - existing-test-3: PASS/FAIL
Result: X/Y passed (previously Y/Y)

评分器类型

1. 基于代码的评分器

使用代码进行确定性检查：

# Check if file contains expected pattern
grep -q "export function handleAuth" src/auth.ts && echo "PASS" || echo "FAIL"

# Check if tests pass
npm test -- --testPathPattern="auth" && echo "PASS" || echo "FAIL"

# Check if build succeeds
npm run build && echo "PASS" || echo "FAIL"

2. 基于模型的评分器

使用 Claude 评估开放式输出：

[MODEL GRADER PROMPT]
Evaluate the following code change:
1. Does it solve the stated problem?
2. Is it well-structured?
3. Are edge cases handled?
4. Is error handling appropriate?

Score: 1-5 (1=poor, 5=excellent)
Reasoning: [explanation]

3. 人工评分器

标记需要人工审查：

[HUMAN REVIEW REQUIRED]
Change: Description of what changed
Reason: Why human review is needed
Risk Level: LOW/MEDIUM/HIGH

指标

pass@k

"k 次尝试中至少一次成功"

• pass@1：首次尝试成功率
• pass@3：3 次尝试内成功
• 典型目标：pass@3 > 90%

pass^k

"k 次试验全部成功"

• 更高的可靠性标准
• pass^3：连续 3 次成功
• 用于关键路径

评估工作流

1. 定义（编码前）

## EVAL DEFINITION: feature-xyz

### Capability Evals
1. Can create new user account
2. Can validate email format
3. Can hash password securely

### Regression Evals
1. Existing login still works
2. Session management unchanged
3. Logout flow intact

### Success Metrics
- pass@3 > 90% for capability evals
- pass^3 = 100% for regression evals

2. 实现

编写代码以通过定义的评估。

3. 评估

# Run capability evals
[Run each capability eval, record PASS/FAIL]

# Run regression evals
npm test -- --testPathPattern="existing"

# Generate report

4. 报告

EVAL REPORT: feature-xyz
========================

Capability Evals:
  create-user:     PASS (pass@1)
  validate-email:  PASS (pass@2)
  hash-password:   PASS (pass@1)
  Overall:         3/3 passed

Regression Evals:
  login-flow:      PASS
  session-mgmt:    PASS
  logout-flow:     PASS
  Overall:         3/3 passed

Metrics:
  pass@1: 67% (2/3)
  pass@3: 100% (3/3)

Status: READY FOR REVIEW

集成模式

实现前

/eval define feature-name

在 .claude/evals/feature-name.md 创建评估定义文件

实现中

/eval check feature-name

运行当前评估并报告状态

实现后

/eval report feature-name

生成完整评估报告

评估存储

在项目中存储评估：

.claude/
  evals/
    feature-xyz.md      # Eval definition
    feature-xyz.log     # Eval run history
    baseline.json       # Regression baselines

最佳实践

1. 编码前定义评估 - 强制清晰思考成功标准
2. 频繁运行评估 - 尽早发现回归
3. 跟踪 pass@k 随时间变化 - 监控可靠性趋势
4. 尽可能使用代码评分器 - 确定性 > 概率性
5. 安全相关需人工审查 - 永远不要完全自动化安全检查
6. 保持评估快速 - 慢评估不会被运行
7. 评估与代码一起版本化 - 评估是一等公民

示例：添加认证

## EVAL: add-authentication

### Phase 1: Define (10 min)
Capability Evals:
- [ ] User can register with email/password
- [ ] User can login with valid credentials
- [ ] Invalid credentials are rejected
- [ ] Session token is generated on login
- [ ] Protected routes require authentication

Regression Evals:
- [ ] Public routes still accessible
- [ ] Existing API endpoints unchanged
- [ ] Database migrations are reversible