checkpoint（checkpoint 搜索）

[IMPORTANT] Use TaskCreate to break ALL work into small tasks BEFORE starting — including tasks for each file read. This prevents context loss from long files. For simple tasks, AI MUST ATTENTION ask user whether to skip.

Critical Thinking Mindset — Apply critical thinking, sequential thinking. Every claim needs traced proof, confidence >80% to act. Anti-hallucination: Never present guess as fact — cite sources for every claim, admit uncertainty freely, self-check output for errors, cross-reference independently, stay skeptical of own confidence — certainty without evidence root of all hallucination.

AI Mistake Prevention — Failure modes to avoid on every task: - Check downstream references before deleting. Deleting components causes documentation and code staleness cascades. Map all referencing files before removal. - Verify AI-generated content against actual code. AI hallucinates APIs, class names, and method signatures. Always grep to confirm existence before documenting or referencing. - Trace full dependency chain after edits. Changing a definition misses downstream variables and consumers derived from it. Always trace the full chain. - Trace ALL code paths when verifying correctness. Confirming code exists is not confirming it executes. Always trace early exits, error branches, and conditional skips — not just happy path. - When debugging, ask "whose responsibility?" before fixing. Trace whether bug is in caller (wrong data) or callee (wrong handling). Fix at responsible layer — never patch symptom site. - Assume existing values are intentional — ask WHY before changing. Before changing any constant, limit, flag, or pattern: read comments, check git blame, examine surrounding code. - Verify ALL affected outputs, not just the first. Changes touching multiple stacks require verifying EVERY output. One green check is not all green checks. - Holistic-first debugging — resist nearest-attention trap. When investigating any failure, list EVERY precondition first (config, env vars, DB names, endpoints, DI registrations, data preconditions), then verify each against evidence before forming any code-layer hypothesis. - Surgical changes — apply the diff test. Bug fix: every changed line must trace directly to the bug. Don't restyle or improve adjacent code. Enhancement task: implement improvements AND announce them explicitly. - Surface ambiguity before coding — don't pick silently. If request has multiple interpretations, present each with effort estimate and ask. Never assume all-records, file-based, or more complex path.

Quick Summary

Goal: Save current analysis context and progress to an external file for recovery after context loss.

Workflow:

1. Gather Context — Collect task state, findings, files analyzed, decisions made
2. Write Checkpoint — Save structured markdown to plans/reports/checkpoint-{timestamp}-{slug}.md
3. Update Todos — Reflect checkpoint creation in task tracking

Key Rules:

• Save checkpoints every 30-60 minutes during complex tasks
• Include file paths, line numbers, and recovery instructions
• Document decisions with rationale for future reference

Be skeptical. Apply critical thinking, sequential thinking. Every claim needs traced proof, confidence percentages (Idea should be more than 80%).

Save Memory Checkpoint

Save current analysis, findings, and progress to an external memory file to prevent context loss during long-running tasks.

Usage

Use this command when:

• Working on complex multi-step tasks (investigation, planning, implementation)
• Before expected context compaction
• At key milestones during feature development
• After completing significant analysis phases

Checkpoint File Location

Files are saved to: plans/reports/checkpoint-{timestamp}-{slug}.md

Instructions

Create a checkpoint file with the following structure:

Step 1: Determine Checkpoint Location

# Get current date for filename
date +%y%m%d-%H%M

Step 2: Gather Context

Collect and document:

1. Current Task - What are you working on?
2. Key Findings - What have you discovered?
3. Files Analyzed - Which files have been read/modified?
4. Progress Summary - What's completed vs remaining?
5. Important Context - Critical information to preserve
6. Next Steps - What should be done next?
7. Open Questions - Unresolved issues

Step 3: Write Checkpoint File

Create a markdown file at plans/reports/checkpoint-YYMMDD-HHMM-{task-slug}.md with:

# Memory Checkpoint: [Task Description]

> Checkpoint created to preserve analysis context during [task type].

## Session Info

- **Created:** [timestamp]
- **Task:** [description]
- **Branch:** [git branch]
- **Phase:** [current phase]

## Current Task Summary

[Brief description of what you're working on]

## Key Findings

### Analysis Results

- [Finding 1]
- [Finding 2]
- [Finding N]

### Patterns Discovered

- [Pattern 1]
- [Pattern 2]

### Dependencies Identified

- [Dependency 1]
- [Dependency 2]

## Files Context

### Analyzed Files

| File            | Purpose   | Relevance       |
| --------------- | --------- | --------------- |
| path/to/file.cs | [purpose] | High/Medium/Low |

### Modified Files

- `path/to/modified.ts` - [change description]

### Pending Files

- `path/to/pending.cs` - [why pending]

## Progress Summary

### Completed

- [x] [Completed item 1]
- [x] [Completed item 2]

### In Progress

- [ ] [Current item]

### Remaining

- [ ] [Remaining item 1]
- [ ] [Remaining item 2]

## Important Context

### Critical Information

[Information that must not be lost]

### Assumptions Made

- [Assumption 1]
- [Assumption 2]

### Decisions Made

- [Decision 1] - [rationale]
- [Decision 2] - [rationale]

## Next Steps

1. [Immediate next action]
2. [Following action]
3. [Subsequent action]

## Open Questions

- [ ] [Question 1]
- [ ] [Question 2]

## Recovery Instructions

To resume this task after context reset:

1. Read this checkpoint file
2. Review [specific files] for context
3. Continue from [specific point]

---

_Checkpoint saved by Claude Code at [timestamp]_

Step 4: Update Todo List

Update your todo list to reflect checkpoint was created:

- [x] Create memory checkpoint at [timestamp]

Best Practices

1. Save checkpoints frequently - Every 30-60 minutes during complex tasks
2. Be specific - Include file paths, line numbers, exact findings
3. Document decisions - Record why choices were made
4. Link related files - Reference other analysis documents
5. Include recovery steps - Make resumption easy

Related Commands

• /context - Load project context
• /compact - Manually trigger context compaction
• /watzup - Generate progress summary

Closing Reminders

• MANDATORY IMPORTANT MUST ATTENTION break work into small todo tasks using TaskCreate BEFORE starting
• MANDATORY IMPORTANT MUST ATTENTION search codebase for 3+ similar patterns before creating new code
• MANDATORY IMPORTANT MUST ATTENTION cite file:line evidence for every claim (confidence >80% to act)
• MANDATORY IMPORTANT MUST ATTENTION add a final review todo task to verify work quality
• MUST ATTENTION apply critical thinking — every claim needs traced proof, confidence >80% to act. Anti-hallucination: never present guess as fact.
• MUST ATTENTION apply AI mistake prevention — holistic-first debugging, fix at responsible layer, surface ambiguity before coding, re-read files after compaction.

[TASK-PLANNING] Before acting, analyze task scope and systematically break it into small todo tasks and sub-tasks using TaskCreate.