Toowiredd/claude-skills-automation
Fully automated memory and context management for Claude Code using hooks - Zero friction, zero context loss
Deep Analysis
全自动的Claude Code记忆和上下文管理系统,通过hooks实现零摩擦的会话恢复、决策跟踪和永久记忆
Core Features
Technical Implementation
- 零手动操作 - 完全消除"hi-ai"、"remember"等手动记忆管理命令
- 即时生产力 - 会话启动即恢复上下文,无缝延续工作
- 神经多样性友好 - 外部化记忆系统补偿SDAM等认知差异
- 企业级集成 - 与Pieces、Copilot等付费订阅工具打通
- 完整可追溯 - 审计日志记录所有自动化决策
- ADHD用户消除上下文切换摩擦,保持心流状态
- SDAM用户建立完整外部记忆系统
- 长期项目维护完整决策历史和上下文
- 团队协作时共享决策记录和阻塞因素
- 企业用户集成付费订阅工具形成完整工作流
- 依赖Claude Code的hook机制,不支持其他Claude环境
- 自动决策提取基于关键词模式,可能误判或遗漏
- 完整transcript保存会占用较大存储空间
- 付费工具集成需要额外订阅成本
- 缺少多用户和团队协作功能
- 没有云同步,仅本地存储
Claude Skills Automation
Fully automated memory and context management for Claude Code
Zero-friction development with automatic context restoration, decision tracking, and permanent memory.
🎯 What This Does
Eliminates all manual memory management in Claude Code through automated hooks:
Before ❌
You: [Start Claude Code]
Claude: "How can I help?"
You: "hi-ai, what were we working on?"
You: "Remember we're using PostgreSQL"
You: "What did we decide about React?"
After ✅
You: [Start Claude Code]
Claude: # Session Context Restored
**Recent Decisions:**
• Using PostgreSQL for database (2 days ago)
• Chose React + Vite for frontend (3 days ago)
• Implementing JWT authentication (5 days ago)
**Memory system active**. Ready to continue!
You: [Just start working - context already there]
Zero manual "hi-ai". Zero manual "remember". Just works.
✨ Features
🚀 Automatic Context Restoration
- Session starts → Context automatically injected
- No more "hi-ai" commands
- Instant continuation from where you left off
💾 Automatic Decision Extraction
- Detects decisions from conversation patterns
- Extracts: "using", "chose", "decided", "implemented"
- Saves to memory automatically
🚧 Automatic Blocker Tracking
- Detects obstacles: "can't", "blocked by", "waiting for"
- Tracks active blockers
- Surfaces them in next session
📝 Automatic Session Saving
- Saves state on session end
- Backs up full transcript
- Never lose context
🔍 File Change Tracking
- Monitors significant file edits
- Logs DECISIONS.md, README, code changes
- Creates audit trail
💪 Neurodivergent-Optimized
- ADHD: Zero friction, immediate execution
- SDAM: Complete external memory system
- Aphantasia: Visual outputs, concrete examples
- Dyschronometria: Time-anchored context
- Anendophasia: Externalized reasoning
🚀 Quick Start
Basic Installation (2 minutes)
# Clone repository
git clone https://github.com/toowiredd/claude-skills-automation
cd claude-skills-automation
# Run basic installer
bash scripts/install.sh
Supercharged Installation with Paid Subscriptions Integration
If you have Pieces.app, GitHub Copilot Pro, Docker Pro, Codegen-ai, Jules CLI, or Codacy:
# Run enhanced installer
bash scripts/install-integrations.sh
# See full integration guide
cat docs/SUBSCRIPTIONS_INTEGRATION.md
That's it! Your next Claude Code session will have full automation.
Verification
# Start Claude Code session
# Context should appear automatically
# Check automation logs
tail -f ~/.claude-memories/automation.log
📋 What Gets Installed
Automation Hooks
Core Hooks (5) - Memory & context automation:
| Hook | Size | Executes When | Performance |
|---|---|---|---|
session-start.sh |
3.1 KB | Session starts | 100ms |
session-end.sh |
2.0 KB | Session ends | 50ms |
stop-extract-memories.sh |
2.6 KB | After each response | 200ms |
post-tool-track.sh |
1.4 KB | After file edits | 10ms |
pre-compact-backup.sh |
1.5 KB | Before compression | 100ms |
Integration Hooks (7) - 🆕 Paid subscriptions:
async-task-jules.sh- Jules CLI async taskserror-lookup-copilot.sh- GitHub Copilot error searchpre-commit-copilot-review.sh- AI code reviewpre-commit-codacy-check.sh- Quality gatespost-tool-save-to-pieces.sh- Pieces.app synctesting-docker-isolation.sh- Docker isolated testscodegen-agent-trigger.sh- Autonomous agents
Total overhead per session: <500ms (core only)
8 Claude Skills
Core Skills (5):
- session-launcher - Zero context loss restoration
- context-manager - Permanent memory management
- error-debugger - Smart debugging with past solutions
- testing-builder - Automatic test generation
- rapid-prototyper - Fast idea validation
New Skills (3) 🆕:
- browser-app-creator - Generates single-file HTML/CSS/JS apps (ADHD-optimized, 60px+ buttons, localStorage)
- repository-analyzer - Analyzes codebases, generates documentation, detects patterns, extracts TODOs
- api-integration-builder - Creates TypeScript API clients with retry logic, rate limiting, and auth
Memory Storage
~/.claude-memories/
├── index.json # Master memory index
├── decisions/ # Architecture decisions
├── blockers/ # Active obstacles
├── procedures/ # How-to guides
├── backups/ # Session backups
├── automation.log # Hook execution log
└── auto-extracted.log # Auto-extracted memories
📖 Documentation
- Quick Start - Get started in 5 minutes
- Complete Guide - Full documentation (3,500+ lines)
- Subscriptions Integration - 🆕 Paid subscriptions integration guide
- Implementation - Technical details
- Research - How we built this
🎓 How It Works
System Architecture
┌─────────────────────────────────────────┐
│ SessionStart Hook (100ms) │
│ • Load memory index │
│ • Extract recent decisions │
│ • Find active blockers │
│ • Inject context into conversation │
└────────────────┬────────────────────────┘
↓
┌─────────────────────────────────────────┐
│ You Work Normally │
│ • Code, debug, design │
│ • Skills activate as needed │
└────────────────┬────────────────────────┘
↓
┌─────────────────────────────────────────┐
│ PostToolUse Hook (10ms) │
│ • Track file modifications │
│ • Log significant changes │
└────────────────┬────────────────────────┘
↓
┌─────────────────────────────────────────┐
│ Stop Hook (200ms) │
│ • Analyze conversation │
│ • Extract decision patterns │
│ • Extract blocker patterns │
└────────────────┬────────────────────────┘
↓
┌─────────────────────────────────────────┐
│ SessionEnd Hook (50ms) │
│ • Save session state │
│ • Backup transcript │
└─────────────────────────────────────────┘
Decision Extraction Patterns
Automatically detects and extracts:
| Pattern | Example | Extracted As |
|---|---|---|
| "using X" | "We're using PostgreSQL" | DECISION |
| "chose X" | "Chose React over Vue" | DECISION |
| "decided X" | "Decided to implement caching" | DECISION |
| "can't X" | "Can't access the API" | BLOCKER |
| "blocked by X" | "Blocked by missing keys" | BLOCKER |
🧪 Example Session
First Session
You: "Let's build a task tracker with React and PostgreSQL"
Claude: "Great! Let's start..."
# Behind the scenes:
# → Stop hook extracts: "build a task tracker with React and PostgreSQL"
# → Logs as DECISION
# → SessionEnd saves state
Next Session
[You start Claude Code]
Claude: # Session Context Restored
**Recent Decisions:**
• Build a task tracker with React and PostgreSQL (1 day ago)
**Last Working On:** Building task tracker
Ready to continue!
# Zero manual effort - context just there
🔧 Requirements
- Claude Code (latest version)
- Bash (pre-installed on Linux/macOS)
- jq (for JSON processing)
# Install jq if needed sudo apt install jq # Ubuntu/Debian brew install jq # macOS
📊 Performance
| Metric | Value |
|---|---|
| Installation time | 2 minutes |
| Hook overhead/session | <500ms |
| Disk usage/year | <100MB |
| Maintenance required | Zero |
🌟 Community & Credits
Built on research from:
Why We Built This
Designed specifically for neurodivergent developers:
- ADHD → Zero friction, immediate results
- SDAM → Complete external memory
- Dyschronometria → Time-anchored context
- Aphantasia → Visual, concrete outputs
🤝 Contributing
We welcome contributions! See CONTRIBUTING.md for guidelines.
Ideas for Contributions
- 🎯 New extraction patterns
- 🔧 Additional hooks
- 📚 More skills
- 🌍 Translations
- 🎨 Better documentation
- 🧪 Tests
📝 Advanced Usage
Custom Decision Patterns
Edit ~/.config/claude-code/hooks/stop-extract-memories.sh:
# Add your own patterns
local decision_patterns="using|chose|decided|MY_PATTERN"
Pieces.app Integration
Three integration options:
- Auto-save code to Pieces
- Export memories to Pieces
- Use Pieces API as backend
Project-Specific Hooks
Create .claude/hooks/ in your project for custom automation.
🐛 Troubleshooting
Hooks Not Firing?
# Check hooks are executable
ls -l ~/.config/claude-code/hooks/*.sh
# Fix permissions
chmod +x ~/.config/claude-code/hooks/*.sh
Context Not Appearing?
# Verify memory index
cat ~/.claude-memories/index.json
jq '.total_memories' ~/.claude-memories/index.json
More help: Troubleshooting Guide
Related Skills
wshobson/agents
wshobsonIntelligent automation and multi-agent orchestration for Claude Code
The most comprehensive Claude Code plugin ecosystem, covering full-stack development scenarios with a three-tier model strategy balancing performance and cost.
ComposioHQ/awesome-claude-skills
ComposioHQA curated list of awesome Claude Skills, resources, and tools for customizing Claude AI workflows
The most comprehensive Claude Skills resource list; connect-apps is a killer feature.
code-yeongyu/oh-my-opencode
code-yeongyuThe Best Agent Harness. Meet Sisyphus: The Batteries-Included Agent that codes like you.
Powerful multi-agent coding tool, but note OAuth limitations.
nextlevelbuilder/ui-ux-pro-max-skill
nextlevelbuilderAn AI SKILL that provide design intelligence for building professional UI/UX multiple platforms
Essential for designers; comprehensive UI/UX knowledge base.
thedotmack/claude-mem
thedotmackA Claude Code plugin that automatically captures everything Claude does during your coding sessions, compresses it with AI (using Claude's agent-sdk), and injects relevant context back into future sessions.
A practical solution for Claude's memory issues.
OthmanAdi/planning-with-files
OthmanAdiClaude Code skill implementing Manus-style persistent markdown planning — the workflow pattern behind the $2B acquisition.
Context engineering best practices; an open-source implementation of Manus mode.

