• User Persona: Define your background, expertise, interests, and communication preferences
• AI Persona: Customize the AI's role, communication style, and interaction mode

📝 Smart Note Retrieval

• Automatically index your Markdown notes
• Intelligently retrieve relevant history based on conversation content
• Naturally reference your past ideas and notes in the dialogue

💬 Personalized Conversation

• Generate personalized responses based on your persona and notes
• Maintain contextual coherence in the conversation
• Reference your ideas like a friend, not mechanically "according to the record"

Usage Scenarios

When you need:

• Personalized communication, not generic replies
• Context-aware responses, with AI remembering your background
• AI to remember and reference your previous thoughts and notes
• A continuous conversation experience, not a fresh start every time

Installation & Usage

Install the Skill

Copy this project into your working directory under the .claude/skills/ folder:

<你的项目根目录>/
└── .claude/
    └── skills/
        └── ai-partner-chat/    # 本技能包
            ├── assets/
            ├── scripts/
            ├── SKILL.md
            └── README.md

Using Skills

In Claude Code, send the following command to activate this skill:

遵循 ai-partner-chat 对话

The AI agent will automatically:

• Read skill configurations and instructions
• Create necessary directory structures (notes/, config/, vector_db/, etc.)
• Initialize according to your needs

Initialization Process

#### Method 1: Let the AI automatically create and configure

For first-time use, simply tell the AI:

我刚刚在 notes 里放入了对应的笔记，请根据笔记内容，进行向量化；并基于笔记内容，推测并更新 user-persona.md，以及最适合我的 ai-persona.md

The AI agent will:

• Analyze the notes in the notes/ directory
• Intelligently segment notes based on their format and create a vector database
• Infer your background and preferences based on the note content
• Automatically generate and update config/user-persona.md
• Recommend and create config/ai-persona.md based on your characteristics

If you want to define the persona yourself:

• The AI agent will automatically create persona files from templates into the config/ directory
• You can manually edit these files to customize the persona
• Then instruct the AI to perform vectorization

Start Conversation

After configuration, simply send the following each time you use it:

遵循 ai-partner-chat 对话

AI will:

• Read your profile to understand your background
• Retrieve relevant historical notes
• Generate personalized, context-aware responses

Project Structure

Skill Package Structure (located at .claude/skills/ai-partner-chat/)

ai-partner-chat/
├── assets/              # 画像模板
│   ├── user-persona-template.md
│   └── ai-persona-template.md
├── scripts/             # 核心脚本
│   ├── chunk_schema.py
│   ├── vector_indexer.py
│   ├── vector_utils.py
│   └── requirements.txt
├── SKILL.md            # 技能详细文档（AI agent 会读取此文件）
└── README.md           # 本文件

User Data Directory (Located in the Project Root Directory)

The AI agent will create the following structure in your project root directory:

<项目根目录>/
├── notes/              # 你的笔记（由你或 AI agent 创建）
├── config/             # 画像配置（由 AI agent 创建）
│   ├── user-persona.md
│   └── ai-persona.md
├── vector_db/          # 向量数据库（由 AI agent 创建）
└── venv/               # Python 虚拟环境（由 AI agent 创建）

Important: User data and skill packages are separated for easy backup and migration.

Workflow

• Load Profiles: Read user profiles and AI profiles to understand the interaction context
• Retrieve Notes: Retrieve the most relevant notes from the vector database based on user queries
• Build Context: Integrate profile information, relevant notes, and conversation history
• Generate Response: Generate personalized and natural responses based on the context

Highlights

🤖 AI Agent Intelligent Chunking

🎯 Natural Citation

📦 Data Independence

Best Practices

Profile Design

• Be Specific and Clear: Vague profiles lead to generic replies
• Include Examples: Show expected interaction patterns in the AI profile
• Update Regularly: Continuously optimize profiles based on conversation quality

Note Management

• Flexible Format: The system adapts to any note structure
• Rich Content: In-depth notes bring better retrieval results
• Timely Updates: Remember to add new notes to the index

Conversation Experience

• Natural Citation: Cite notes only when truly relevant
• Maintain Flow: Don’t let citations disrupt the natural rhythm of conversation
• Focus on Quality: Prioritize meaningful connections over quantity

Maintenance and Updates

Add New Note

我刚刚在 notes 里添加了新笔记，请更新向量数据库

The AI agent will automatically analyze new notes and update the index.

Update Profile

请根据我最近的笔记内容，更新 user-persona.md 和 ai-persona.md

Rebuild Index

请重新初始化向量数据库

Notes

• First Run: When the AI agent creates the vector database for the first time, it will automatically download the embedding model (about 4.3GB), please be patient
• Python Environment: The AI agent will automatically create a virtual environment and install the required dependencies
• Data Storage: All data (notes, profiles, vector libraries) are stored in the project root directory, not in the skill package directory
• Skill Location: Ensure the skill package is located in the .claude/skills/ai-partner-chat/ directory

More Information

For detailed technical documentation and usage instructions, please refer to the SKILL.md file.

Let AI become a conversation partner who truly understands you, not just a tool.