SuperAPI AI Directory

memory-checkpoint

General & other

View repository

Updated 10-16 09:44

About this skill

Memory Checkpoint Skill (記憶檢查點)

Description

Proactively externalize detailed memory to Memory Bank before the conversation is compressed by Summarize, to avoid losing important context.

Trigger Conditions

  • "Memory Checkpoint", "checkpoint", "save"
  • "save memory", "externalize memory"
  • 「sync memory」、「dump context」
  • Proactively trigger when the conversation is long (recommended every 10-15 turns or after major progress)

⚠️ Why is this Skill needed?

When the conversation becomes too long, the system automatically Summarizes the conversation history, which may cause:

  • Detailed code change records to be omitted
  • The discussion context behind decisions to be lost
  • File paths and concrete implementation details to become vague
  • Time to be spent rebuilding context

Solution: Before being Summarized, proactively write key memories to the memory-bank/ directory.


📋 Checkpoint Contents

1️⃣ activeContext.md - Current work focus

## 當前工作焦點
<!-- 當前正在做什麼 -->

## 進行中的變更
<!-- 具體的檔案和修改 -->

## 待處理事項
<!-- 下一步要做什麼 -->

## 關鍵決策
<!-- 本次對話做的重要決定 -->

## 相關檔案
<!-- 涉及的檔案路徑列表 -->

2️⃣ progress.md - Progress tracking

## Done (已完成)
- [x] 具體完成的事項(含檔案路徑)

## Doing (進行中)
- [ ] 正在進行的工作

## Next (下一步)
- [ ] 計劃要做的事項

3️⃣ decisionLog.md - Decision log

## YYYY-MM-DD

### 決策:[決策標題]
- **背景**:為什麼需要這個決定
- **選項**:考慮過的方案
- **決定**:最終選擇
- **原因**:選擇的理由

4️⃣ architect.md - Architecture record (if changed)

## 架構變更記錄

### YYYY-MM-DD: [變更標題]
- **變更內容**:
- **影響範圍**:
- **相關檔案**:

🚀 Checkpoint Execution Steps

Automatic execution (recommended)

When the following situations are detected, the AI should proactively perform a Checkpoint:

  1. Conversation length indicators

    • Conversation exceeds 10 turns
    • Cumulative modifications exceed 5 files
    • Completion of an important feature/fix
  2. Work phase transition

    • Transition from design phase to implementation
    • Completion of a PR or Commit
    • Switching to a different feature area
  3. Explicit indicators

    • User says "note this down first", "checkpoint"
    • User says "continue later", "I'm going to be busy for a while"

Manual execution

The user can say at any time:

  • "Memory Checkpoint"
  • "sync memory bank"
  • "save current progress"

📝 Checkpoint Template

Quick Checkpoint

Update activeContext.md:

# Active Context

> Last checkpoint: 2024-XX-XX HH:MM

## 🎯 當前焦點
[一句話描述當前在做什麼]

## 📝 本次對話摘要
1. [做了什麼 1]
2. [做了什麼 2]
3. [做了什麼 3]

## 📁 變更的檔案
- `path/to/file1.py` - [變更摘要]
- `path/to/file2.ts` - [變更摘要]

## ⚠️ 待解決
- [ ] 問題 1
- [ ] 問題 2

## 💡 重要決定
- 決定 A:[原因]
- 決定 B:[原因]

## 🔜 下一步
1. [下一步 1]
2. [下一步 2]

Full Checkpoint

Update all relevant Memory Bank files:

memory-bank/
├── activeContext.md  ← 當前焦點 + 變更檔案
├── progress.md       ← Done/Doing/Next
├── decisionLog.md    ← 新決策(如有)
├── architect.md      ← 架構變更(如有)
├── productContext.md ← 技術棧變更(如有)
└── systemPatterns.md ← 新模式/慣例(如有)

🔄 Integration with Other Skills

Use with git-precommit

git commit 前 → 觸發 memory-checkpoint → 同步 Memory Bank → commit

Use with memory-updater

memory-checkpoint = 批次更新
memory-updater = 增量更新

💬 Conversation Examples

User-triggered

User: checkpoint

AI: 📸 執行記憶檢查點...

已更新 Memory Bank:
- ✅ activeContext.md - 記錄當前焦點與 5 個變更檔案
- ✅ progress.md - 新增 3 個已完成項目
- ✅ decisionLog.md - 記錄 2 個決策

下次對話可從此處繼續,即使被 Summarize 也不會遺失關鍵資訊。

AI proactively triggered

AI: 💡 偵測到對話較長且有多項進展,建議執行記憶檢查點。
    是否要我同步 Memory Bank?

User: 好

AI: 📸 執行記憶檢查點...
    [更新 Memory Bank]

⚙️ Configuration Options

In .vscode/settings.json configurable:

{
  "claude.skills.memoryCheckpoint": {
    "autoTrigger": true,
    "triggerThreshold": {
      "conversationTurns": 10,
      "filesModified": 5
    },
    "includeFiles": [
      "activeContext.md",
      "progress.md",
      "decisionLog.md"
    ]
  }
}

📊 Checkpoint Quality Check

A good Checkpoint should include:

Item Required Description
Current focus ✅ One sentence describing what is being done
Changed file list ✅ Full path + brief description
Pending items ✅ Work not yet completed
Important decisions ⚪ If there are new decisions
Next steps ✅ What to do next
Timestamp ✅ Know when this state was from

🛠️ Implementation Tips

Instructions for the AI

In CLAUDE.md or AGENTS.md add:

## Memory Checkpoint 規則

1. **主動觸發時機**
   - 對話超過 10 輪時,主動建議 checkpoint
   - 完成重大功能後,主動執行 checkpoint
   - 使用者說要離開時,主動執行 checkpoint

2. **Checkpoint 內容**
   - 必須包含:具體檔案路徑、變更摘要、下一步
   - 避免:過於籠統的描述、遺漏重要細節

3. **格式要求**
   - 使用時間戳記
   - 檔案路徑使用相對路徑
   - 保持簡潔但完整