Master backend architectures, REST API design, microservices, and server-side development across multiple languages. Use this as a central index to access specialized sub-skills.
한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.
[72] CLOSE. Save and restore session context between conversations. Use when ending a session to preserve progress, or starting a new session to restore context. Triggers on "save session", "end session", "preserve context", "handoff", "continue from last time", or when context window is running low.
한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.
설치 방법 보기설치하기 전에 에이전트에게 전달되는 지시 내용을 확인할 수 있습니다.
Core Principle: Never lose progress. Save context before ending, restore context when starting.
This skill handles two modes:
SAVE triggers:
RESTORE triggers:
Before saving, collect data from all sources:
| Source | What to Read | Purpose |
|---|---|---|
.sessions/SESSION_[date]-[name].md | Progress Log / Findings / Decisions | Raw data for lessons |
| Chat history | Current conversation context | What was discussed, tried, decided |
MEMORY.md | Lessons Inbox | Existing lessons |
| Files touched | Recent changes | What was modified |
Critical: Don't start saving until you've read all sources!
Review the session log Progress Log and extract lessons:
MEMORY.md Lessons (Inbox)Progress Log Analysis:
❌ Failed attempts → Why did they fail? → Principle for future
✅ Successful fixes → What made it work? → Confirm understanding
🔄 Experiments → What was learned? → Document insight
What was accomplished:
Key choices made and why:
Clear continuation path:
Important state to remember:
💾 **SESSION-SAVE 150**
**Date:** [YYYY-MM-DD HH:MM]
**Project:** [Project name/path]
## Progress Made
- ✅ [Completed item 1]
- ✅ [Completed item 2]
- 🔄 [In progress item]
## Current State
- Working on: [Current focus]
- Files touched: [Key files]
- Status: [Overall status]
## Key Decisions
| Decision | Reasoning |
|----------|-----------|
| [Choice 1] | [Why] |
| [Choice 2] | [Why] |
## Next Steps (Priority Order)
1. **Immediate:** [Next action]
2. **Then:** [Following action]
3. **Later:** [Future action]
## Blockers & Risks
- ⚠️ [Blocker/Risk 1]
- ⚠️ [Blocker/Risk 2]
## Context to Remember
- [Important detail 1]
- [Important detail 2]
---
**Handoff Status:** ✅ Ready for next session
🔄 **SESSION-RESTORE 150**
**Restoring from:** [Date of saved session]
## Previous Progress
[Summary of what was done]
## Continuing From
- Last state: [Where we left off]
- Next step: [What to do now]
## Context Restored
- ✅ [Verified context 1]
- ✅ [Verified context 2]
- ⚠️ [Needs verification]
## Ready to Continue
Starting with: [First action]
For fast saves when time is short:
💾 **QUICK SAVE**
**Done:** [What was accomplished]
**Current:** [Where we are]
**Next:** [What to do next]
**Remember:** [Key context]
If 74-mid-session-save-150 was used during the session:
Session Flow:
┌─────────────────────────────────────────────────────────┐
│ mid-session-save → mid-session-save → close-session │
│ ↓ ↓ ↓ │
│ Progress Log #1 Progress Log #2 READ ALL LOGS │
│ (raw data) (more data) EXTRACT LESSONS │
│ FINAL HANDOFF │
└─────────────────────────────────────────────────────────┘
Close-session responsibility:
.sessions/SESSION_[date]-[name].mdMEMORY.md Lessons (Inbox)📁 SAVE LOCATIONS
├── MEMORY.md # Project memory file + extracted lessons
├── .sessions/SESSION_[date]-[name].md # Session-specific log (read & update)
├── Project README # For major milestones
└── Git commit message # For code changes
| Location | READ | WRITE |
|---|---|---|
.sessions/SESSION_[date]-[name].md | Progress Log, checkpoints, findings | Final handoff state |
MEMORY.md | Lessons Inbox | New lessons |
| Chat history | Full conversation context | — |
| Files touched | What was modified | — |
Input: Progress Log from .sessions/SESSION_[date]-[name].md
### Checkpoint #1
❌ Didn't Work: Direct state update in useEffect — infinite loop
Reason: dependency array included computed object
### Checkpoint #2
🔄 Tried: useMemo for computed dependency
✅ Worked: useMemo solved the loop — stable reference now
### Checkpoint #3
❌ Didn't Work: Hardcoded colors in component
Reason: Violates project convention (use themeColors)
✅ Worked: Imported from @/utils/themeColors
Output: Lessons extracted to MEMORY.md
### 2026-01-10 useEffect dependency stability
**Problem:** Infinite loop from computed object in useEffect deps
**Attempts:** Direct state update failed
**Solution:** Use useMemo to create stable reference
**Why it worked:** useMemo memoizes the computed value, preventing new reference on each render
**Principle:** Never use computed objects in useEffect deps — wrap with useMemo or extract primitive IDs
### 2026-01-10 Color convention adherence
**Problem:** Hardcoded colors violated project standards
**Solution:** Import from @/utils/themeColors
**Principle:** All colors must come from central theme utilities, never hardcoded
"Was working on stuff. Continue later."
Result: Next session spends 30 min figuring out what "stuff" was
💾 SESSION-SAVE 150
Progress Made:
- ✅ Created 7 skills from protocols
- ✅ Refactored chain-flow to include action-plan
- 🔄 Working through remaining protocols
Current State:
- Converting PROTOCOLS_GENERALIZED.md to skills
- 7/13 protocols converted
Key Decisions:
| Decision | Reasoning |
|----------|-----------|
| Renamed "Plan-First" → "chain-flow" | It's an orchestrator, not just a plan |
| Separated action-plan as own skill | Needed as component in chains |
Next Steps:
1. **Immediate:** Create a mid-session checkpoint (`74-mid-session-save-150`) if context is low
2. **Then:** Continue with remaining protocols
3. **Later:** Test skills integration
Context to Remember:
- Skills go in ./.codex/skills/ folder
- Each skill needs SKILL.md with yaml frontmatter
- chain-flow orchestrates other skills
Handoff Status: ✅ Ready
| Failure | Detection | Recovery |
|---|---|---|
| No save | Session ended without handoff | Reconstruct from memory, files, git |
| Vague save | Can't understand notes | Ask questions, check artifacts |
| Outdated context | Things changed since save | Verify current state, update |
| Missing decisions | Don't know why choices made | Review code/docs, make new decision |
| Skipped Progress Log | Lessons not extracted | Re-read session log, extract |
| Didn't read sources | Incomplete handoff | Gather from all sources, redo save |
Remember: Future you (or next session) will thank present you for good notes. Context loss is expensive — prevention is cheap.
아직 리뷰가 없습니다. 사용해 본 소감을 남겨 주세요.
로그인 없이 쓸 수 있습니다. 사용 후기, 질문, 작동하지 않은 환경 등을 편하게 남겨 주세요.
아직 댓글이 없습니다. 첫 댓글을 남겨 보세요.
Master backend architectures, REST API design, microservices, and server-side development across multiple languages. Use this as a central index to access specialized sub-skills.
한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.
在每次回复中使用固定前缀并列出本次实际参考的指令/文件,便于校验是否遵循上下文与规则。
한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.
Automate 21risk tasks via Rube MCP (Composio). Always search tools first for current schemas.
한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.
Automate 2chat tasks via Rube MCP (Composio). Always search tools first for current schemas.
한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.
Arquitecto de Soluciones Principal y Consultor Tecnológico de Andru.ia. Diagnostica y traza la hoja de ruta óptima para proyectos de IA en español.
한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.
[00] META. Orchestrate skills into dynamic chains for complex tasks. Analyzes the task, discovers available skills, builds an optimal chain, explains why each skill is needed, and executes step-by-step with user confirmation. Use for any complex task requiring multiple thinking/research/analysis steps. Triggers on "plan this", "how to approach", "what's the strategy", "build a plan", or any multi-step problem requiring skill orchestration.
한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.