compound-learnings

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Compound Learnings

整合学习成果

Transform ephemeral session learnings into permanent, compounding capabilities.

将临时的会话学习成果转化为可不断积累的永久能力。

When to Use

适用场景

"What should I learn from recent sessions?"
"Improve my setup based on recent work"
"Turn learnings into skills/rules"
"What patterns should become permanent?"
"Compound my learnings"

“我应该从近期会话中学到什么？”
“根据近期工作改进我的设置”
“将学习成果转化为Skills/规则”
“哪些模式应该固化为永久规范？”
“整合我的学习成果”

Process

操作流程

Step 1: Gather Learnings

步骤1：收集学习成果

bash

undefined

bash

undefined

List learnings (most recent first)

列出学习成果（按时间倒序）

ls -t $CLAUDE_PROJECT_DIR/.claude/cache/learnings/*.md | head -20

Count total

统计总数

ls $CLAUDE_PROJECT_DIR/.claude/cache/learnings/*.md | wc -l


Read the most recent 5-10 files (or specify a date range).

ls $CLAUDE_PROJECT_DIR/.claude/cache/learnings/*.md | wc -l


读取最近的5-10个文件（或指定日期范围）。

Step 2: Extract Patterns (Structured)

步骤2：提取结构化模式

For each learnings file, extract entries from these specific sections:

Section Header	What to Extract
`## Patterns` or `Reusable techniques`	Direct candidates for rules
`Takeaway:` or `Actionable takeaway:`	Decision heuristics
`## What Worked`	Success patterns
`## What Failed`	Anti-patterns (invert to rules)
`## Key Decisions`	Design principles

Build a frequency table as you go:

markdown

| Pattern | Sessions | Category |
|---------|----------|----------|
| "Check artifacts before editing" | abc, def, ghi | debugging |
| "Pass IDs explicitly" | abc, def, ghi, jkl | reliability |

针对每个学习成果文件，从以下特定章节提取内容：

章节标题	提取内容
`## Patterns` 或 `Reusable techniques`	可直接转化为规则的候选内容
`Takeaway:` 或 `Actionable takeaway:`	决策启发式规则
`## What Worked`	成功模式
`## What Failed`	反模式（可转化为规则）
`## Key Decisions`	设计原则

在提取过程中构建频率表：

markdown

| 模式 | 会话ID | 分类 |
|---------|----------|----------|
| “编辑前检查产物” | abc, def, ghi | 调试 |
| “显式传递ID” | abc, def, ghi, jkl | 可靠性 |

Step 2b: Consolidate Similar Patterns

步骤2b：合并相似模式

Before counting, merge patterns that express the same principle:

Example consolidation:

"Artifact-first debugging"
"Verify hook output by inspecting files"
"Filesystem-first debugging" → All express: "Observe outputs before editing code"

Use the most general formulation. Update the frequency table.

统计前，合并表达相同原则的模式：

合并示例：

“以产物为核心的调试”
“通过检查文件验证Hook输出”
“以文件系统为核心的调试” → 统一为：“编辑代码前先观察输出结果”

使用最通用的表述，并更新频率表。

Step 3: Detect Meta-Patterns

步骤3：识别元模式

Critical step: Look at what the learnings cluster around.

If >50% of patterns relate to one topic (e.g., "hooks", "tracing", "async"): → That topic may need a dedicated skill rather than multiple rules → One skill compounds better than five rules

Ask yourself: "Is there a skill that would make all these rules unnecessary?"

关键步骤： 分析学习成果的聚类方向。

如果超过50%的模式围绕同一主题（如“Hooks”“追踪”“异步”）： → 该主题可能需要一个专属Skill，而非多条规则 → 一个Skill的积累效果优于五条规则

自问：“是否存在某个Skill可以替代所有这些规则？”

Step 4: Categorize (Decision Tree)

步骤4：分类（决策树）

For each pattern, determine artifact type:

Is it a sequence of commands/steps?
  → YES → SKILL (executable > declarative)
  → NO ↓

Should it run automatically on an event (SessionEnd, PostToolUse, etc.)?
  → YES → HOOK (automatic > manual)
  → NO ↓

Is it "when X, do Y" or "never do X"?
  → YES → RULE
  → NO ↓

Does it enhance an existing agent workflow?
  → YES → AGENT UPDATE
  → NO → Skip (not worth capturing)

Artifact Type Examples:

Pattern	Type	Why
"Run linting before commit"	Hook (PreToolUse)	Automatic gate
"Extract learnings on session end"	Hook (SessionEnd)	Automatic trigger
"Debug hooks step by step"	Skill	Manual sequence
"Always pass IDs explicitly"	Rule	Heuristic

针对每个模式，确定其产物类型：

它是一系列命令/步骤吗？
  → 是 → SKILL（可执行优于声明式）
  → 否 ↓

它是否应在特定事件（SessionEnd、PostToolUse等）触发时自动运行？
  → 是 → HOOK（自动优于手动）
  → 否 ↓

它属于“当X发生时执行Y”或“永远不要执行X”的类型吗？
  → 是 → RULE
  → 否 ↓

它是否能增强现有Agent的工作流？
  → 是 → AGENT 更新
  → 否 → 跳过（不值得固化）

产物类型示例：

模式	类型	原因
“提交前运行代码检查”	Hook（PreToolUse）	自动校验关卡
“会话结束时提取学习成果”	Hook（SessionEnd）	自动触发
“逐步调试Hooks”	Skill	手动执行流程
“始终显式传递ID”	Rule	启发式规则

Step 5: Apply Signal Thresholds

步骤5：应用信号阈值

Occurrences	Action
1	Note but skip (unless critical failure)
2	Consider - present to user
3+	Strong signal - recommend creation
4+	Definitely create

出现次数	操作
1次	记录但跳过（除非是严重故障）
2次	纳入考虑 - 提交给用户确认
3次及以上	强信号 - 建议创建
4次及以上	务必创建

Step 6: Propose Artifacts

步骤6：提交产物提案

Present each proposal in this format:

markdown

---

按照以下格式提交每个提案：

markdown

---

Pattern: [Generalized Name]

模式：[通用名称]

Signal: [N] sessions ([list session IDs])

Category: [debugging / reliability / workflow / etc.]

Artifact Type: Rule / Skill / Agent Update

Rationale: [Why this artifact type, why worth creating]

Draft Content: ```markdown [Actual content that would be written to file] ```

File:

.claude/rules/[name].md

.claude/skills/[name]/SKILL.md


Use `AskUserQuestion` to get approval for each artifact (or batch approval).

信号强度： [N] 次会话（[会话ID列表]）

分类： [调试 / 可靠性 / 工作流 / 其他]

产物类型： Rule / Skill / Agent 更新

理由： [选择该产物类型的原因，以及值得创建的依据]

草案内容： ```markdown [将写入文件的实际内容] ```

文件路径：

.claude/rules/[name].md

或

.claude/skills/[name]/SKILL.md


使用`AskUserQuestion`获取用户对每个产物的批准（或批量批准）。

Step 7: Create Approved Artifacts

步骤7：创建已批准的产物

For Rules:

针对Rules：

bash

undefined

bash

undefined

Write to rules directory

写入rules目录

cat > $CLAUDE_PROJECT_DIR/.claude/rules/<name>.md << 'EOF'

Rule Name

规则名称

[Context: why this rule exists, based on N sessions]

[背景：基于N次会话，该规则的存在原因]

Pattern

模式

[The reusable principle]

[可复用的原则]

DO

应执行

[Concrete action]

[具体操作]

DON'T

禁止执行

[Anti-pattern]

[反模式]

Source Sessions

来源会话

[session-id-1]: [what happened]
[session-id-2]: [what happened] EOF

undefined

undefined

For Skills:

针对Skills：

Create

.claude/skills/<name>/SKILL.md

with:

Frontmatter (name, description, allowed-tools)
When to Use
Step-by-step instructions (executable)
Examples from the learnings

Add triggers to

skill-rules.json

if appropriate.

创建

.claude/skills/<name>/SKILL.md

文件，包含：

前置信息（名称、描述、允许使用的工具）
适用场景
分步操作说明（可执行）
来自学习成果的示例

如需，在

skill-rules.json

中添加触发条件。

For Hooks:

针对Hooks：

Create shell wrapper + TypeScript handler:

bash

undefined

创建Shell包装器 + TypeScript处理器：

bash

undefined

Shell wrapper

Shell包装器

cat > $CLAUDE_PROJECT_DIR/.claude/hooks/<name>.sh << 'EOF' #!/bin/bash set -e cd "$CLAUDE_PROJECT_DIR/.claude/hooks" cat | node dist/<name>.mjs EOF chmod +x $CLAUDE_PROJECT_DIR/.claude/hooks/<name>.sh


Then create `src/<name>.ts`, build with esbuild, and register in `settings.json`:

```json
{
  "hooks": {
    "EventName": [{
      "hooks": [{
        "type": "command",
        "command": "$CLAUDE_PROJECT_DIR/.claude/hooks/<name>.sh"
      }]
    }]
  }
}


然后创建`src/<name>.ts`，使用esbuild构建，并在`settings.json`中注册：

```json
{
  "hooks": {
    "EventName": [{
      "hooks": [{
        "type": "command",
        "command": "$CLAUDE_PROJECT_DIR/.claude/hooks/<name>.sh"
      }]
    }]
  }
}

For Agent Updates:

针对Agent更新：

Edit existing agent in

.claude/agents/<name>.md

to add the learned capability.

编辑

.claude/agents/<name>.md

中的现有Agent，添加新习得的能力。

Step 8: Summary Report

步骤8：总结报告

markdown

undefined

markdown

undefined

Compounding Complete

整合完成

Learnings Analyzed: [N] sessions Patterns Found: [M] Artifacts Created: [K]

分析的学习成果： [N] 次会话 发现的模式： [M] 个 创建的产物： [K] 个

Created:

已创建：

Rule:
```
explicit-identity.md
```
- Pass IDs explicitly across boundaries
Skill:
```
debug-hooks
```
- Hook debugging workflow

Rule:
```
explicit-identity.md
```
- 跨边界显式传递ID
Skill:
```
debug-hooks
```
- Hook调试工作流

Skipped (insufficient signal):

已跳过（信号不足）：

"Pattern X" (1 occurrence)

Your setup is now permanently improved.

undefined

“模式X”（出现1次）

你的设置现已得到永久改进。

undefined

Quality Checks

质量检查

Before creating any artifact:

Is it general enough? Would it apply in other projects?
Is it specific enough? Does it give concrete guidance?
Does it already exist? Check
```
.claude/rules/
```
and
```
.claude/skills/
```
first
Is it the right type? Sequences → skills, heuristics → rules

创建任何产物前，请检查：

通用性足够吗？ 能否应用于其他项目？
具体性足够吗？ 是否提供了明确的指导？
是否已存在？ 先检查
```
.claude/rules/
```
和
```
.claude/skills/
```
目录
类型是否正确？ 步骤序列→Skills，启发式规则→Rules

Files Reference

文件参考

Learnings:
```
.claude/cache/learnings/*.md
```
Skills:
```
.claude/skills/<name>/SKILL.md
```
Rules:
```
.claude/rules/<name>.md
```

Hooks:

.claude/hooks/<name>.sh

src/<name>.ts

dist/<name>.mjs

Agents:
```
.claude/agents/<name>.md
```
Skill triggers:
```
.claude/skills/skill-rules.json
```
Hook registration:
```
.claude/settings.json
```
→
```
hooks
```
section

学习成果：
```
.claude/cache/learnings/*.md
```
Skills：
```
.claude/skills/<name>/SKILL.md
```
规则：
```
.claude/rules/<name>.md
```

Hooks：

.claude/hooks/<name>.sh

src/<name>.ts

dist/<name>.mjs

Agents：
```
.claude/agents/<name>.md
```
Skill触发条件：
```
.claude/skills/skill-rules.json
```
Hook注册：
```
.claude/settings.json
```
→
```
hooks
```
章节