Co-Pilot / 辅助式
更新于 24 days ago

personal-ai-infrastructure

Rrobdtaylor
0.0k
robdtaylor/personal-ai-infrastructure
80
Agent 评分

💡 摘要

个人AI基础设施通过模块化技能和自动化会话管理增强Claude Code。

🎯 适合人群

希望增强其应用程序的AI开发人员需要定制AI工作流程的企业需要多源数据管理的研究人员对个人AI解决方案感兴趣的技术爱好者寻求简化流程的自动化工程师

🤖 AI 吐槽:该项目涉及网络访问和文件系统修改,如果配置不当,可能会暴露敏感数据。实施严格的访问控制并验证输入以降低风险。

安全分析中风险

该项目涉及网络访问和文件系统修改,如果配置不当,可能会暴露敏感数据。实施严格的访问控制并验证输入以降低风险。

Personal AI Infrastructure (PAI)

A skills-based context management system for Claude Code. PAI extends Claude Code with domain-specific knowledge, custom workflows, and automated session management.

What is PAI?

PAI is a personal AI infrastructure layer that sits on top of Claude Code, providing:

  • Skills System - Modular context packages that load relevant knowledge on-demand
  • Hook System - Event-driven automation for session lifecycle management
  • Memory System - Session history, learnings, and signal capture
  • Custom Commands - Slash commands for common workflows

Prerequisites

  • Claude Code CLI installed and authenticated
  • Bun runtime (for hooks and tools)
  • macOS or Linux (Windows via WSL)

Installation

1. Clone the repository

git clone https://github.com/robdtaylor/personal-ai-infrastructure.git ~/.claude

Note: If you already have a ~/.claude directory, back it up first:

mv ~/.claude ~/.claude-backup

2. Copy and configure settings

cd ~/.claude cp settings.json.template settings.json

Edit settings.json to customize:

  • Environment variables
  • Hook configurations
  • Permission settings

3. Install dependencies

# Install Bun if not already installed curl -fsSL https://bun.sh/install | bash # Install hook dependencies cd ~/.claude/hooks && bun install

4. Personalize your setup

Edit the following files in skills/CORE/USER/:

| File | Purpose | |------|---------| | BASICINFO.md | Your contact information | | CONTACTS.md | Directory of people you work with | | ABOUTME.md | Personal background for AI context | | DAIDENTITY.md | AI assistant name and personality | | TECHSTACKPREFERENCES.md | Your preferred technologies |

5. Verify installation

claude

You should see the PAI context load at session start.

Directory Structure

~/.claude/
├── skills/           # Skill packages (context modules)
│   ├── CORE/         # Core identity and system docs
│   ├── Research/     # Multi-source research workflows
│   ├── Browser/      # Web automation
│   └── ...           # Domain-specific skills
├── hooks/            # Event-driven automation
├── commands/         # Custom slash commands
├── bin/              # CLI tools and utilities
├── MEMORY/           # Session history and learnings
└── settings.json     # Claude Code configuration

Skills Included

Manufacturing & Operations

  • AutomotiveManufacturing - IATF 16949, work instructions, SOPs
  • A3CriticalThinking - Toyota-style problem solving
  • CNCSetup - Machine setup sheets
  • CuttingParams - Feeds and speeds reference
  • PlantCapability - Machine shop capabilities
  • QuoteEstimator - Cycle time and cost estimation
  • SkillsMatrix - Operator competency tracking
  • ToolCrib - Tool inventory management
  • TribalKnowledge - Expertise capture
  • MaintenancePM - Preventive maintenance schedules
  • HoshinKanri - Strategy deployment (X-Matrix, catchball)
  • SapEcc - SAP transaction codes and processes
  • SupplyChain - Purchasing and materials management

Business & Strategy

  • BusinessStrategy - Financial analysis, valuations, business plans
  • PowerPlatform - Power Automate flow generation

Infrastructure & Automation

  • HomeAssistant - Smart home integration
  • Infrastructure - Network services and endpoints
  • MobileApp - Mobile web interface for PAI
  • Proxmox - VM management
  • Unifi - Network monitoring
  • n8n - Workflow automation

AI & Research

  • Research - Multi-source parallel research
  • AgentObservability - Real-time agent monitoring dashboard
  • PhotoRealisticArt - AI image generation prompting
  • Art - Excalidraw-style visualizations

Core

  • CORE - Identity, preferences, security protocols
  • Browser - Web automation and verification
  • Context - Obsidian vault integration
  • CreateSkill - Skill authoring tools
  • Upgrades - Track Claude Code updates

Creating Custom Skills

Skills are modular context packages. Create a new skill:

mkdir -p ~/.claude/skills/MySkill

Create SKILL.md:

--- name: MySkill description: Brief description. USE WHEN user says 'keyword1', 'keyword2'. --- # MySkill ## Overview What this skill does... ## Workflows - WorkflowName: Description ## Reference Key information the AI should know...

Regenerate the skill index:

bun run ~/.claude/tools/GenerateSkillIndex.ts

Hooks

PAI uses Claude Code's hook system for automation:

| Hook | Purpose | |------|---------| | SessionStart | Load core context, initialize session | | PreToolUse | Security validation for Bash commands | | PostToolUse | Capture signals and checkpoints | | Stop | Capture learnings and session summary |

Configure hooks in settings.json.

Custom Commands

Slash commands are defined in ~/.claude/commands/. Example:

# /mycommand Description of what this command does. ## Instructions Steps for the AI to follow...

Configuration

Environment Variables

Set in settings.json under env:

{ "env": { "PAI_DIR": "$HOME/.claude", "TIME_ZONE": "Europe/London" } }

Security

The security-validator.ts hook blocks dangerous bash commands. Customize patterns in skills/CORE/USER/PAISECURITYSYSTEM/patterns.yaml.

Troubleshooting

Hooks not running

  1. Check Bun is installed: bun --version
  2. Verify hook paths in settings.json
  3. Check hook file permissions: chmod +x ~/.claude/hooks/*.ts

Skills not loading

  1. Regenerate skill index: bun run ~/.claude/tools/GenerateSkillIndex.ts
  2. Check skill has valid YAML frontmatter in SKILL.md

Context not appearing

  1. Verify load-core-context.ts hook is configured in SessionStart
  2. Check CORE/SKILL.md exists and has valid content

Contributing

  1. Fork this repository
  2. Create a feature branch
  3. Make your changes
  4. Submit a pull request

License

MIT License - See LICENSE file for details.

Acknowledgments

Built for Claude Code by Anthropic.

五维分析
清晰度8/10
创新性8/10
实用性9/10
完整性8/10
可维护性7/10
优缺点分析

优点

  • 模块化技能提供量身定制的AI体验
  • 事件驱动的自动化提高效率
  • 全面的会话管理
  • 支持多个领域和行业

缺点

  • 需要多个依赖项
  • 初学者的设置可能较复杂
  • 对Windows支持有限
  • 定制可能需要技术知识

相关技能

pytorch

S
toolCode Lib / 代码库
92/ 100

“它是深度学习的瑞士军刀,但祝你好运能从47种安装方法里找到那个不会搞崩你系统的那一个。”

agno

S
toolCode Lib / 代码库
90/ 100

“它承诺成为智能体领域的Kubernetes,但得看开发者有没有耐心学习又一个编排层。”

nuxt-skills

S
toolCo-Pilot / 辅助式
90/ 100

“这本质上是一份组织良好的小抄,能把你的 AI 助手变成一只 Nuxt 框架的复读机。”

免责声明:本内容来源于 GitHub 开源项目,仅供展示和评分分析使用。

版权归原作者所有 robdtaylor.