Obsidian Rag 1a1fd3d3

MCP文档知识

提供方 GabrielFersPin

文档知识 stdio · 本地进程 无需鉴权 社区 / 第三方 Claude DesktopClaude
查看接入方式

一个将Obsidian知识库转化为智能学习助手的检索增强生成(RAG)系统,支持本地语义搜索和AI查询。

一句话选型结论

按资料字段整理

Obsidian Rag 1a1fd3d3 属于文档知识类 MCP 服务器,stdio · 本地进程,无需鉴权,可直接接入 Claude Desktop、Claude,有可直接执行的安装命令。接入前先确认权限范围;涉及写入、删除或付款的操作,建议先在测试环境验证。

分类文档知识 传输方式stdio · 本地进程 鉴权无需鉴权 部署方式本地部署
GitHub 星标 2 仓库热度,无公开仓库时显示未标注
工具数 3 暴露给模型的工具数量
资源 / 提示词 0 / 0 资源与提示词模板数量
维护方 社区 / 第三方 核验等级 仅资料核对

工具与能力

文档知识
数量来自来源目录的标注,具体工具清单以官方文档为准
工具数量3暴露给模型可调用的工具个数
资源数量0可读取的资源条目
提示词数量0内置提示词模板数量

能力标签

支持客户端

适用场景

知识管理Python本地部署语义搜索本地AI学习助手笔记检索

安装与配置

stdio · 本地进程 无需鉴权
客户端在本机拉起子进程,数据不出本机
python3 -m venv venv

这条命令只是入口,具体启动参数请按官方文档补齐,不要直接照搬到别的服务器。

  1. 1
    准备运行环境

    stdio 服务在本地起进程,需要先装好 Python,再按安装命令拉取依赖。

  2. 2
    安装服务

    复制上面的安装命令执行;命令里带占位符的位置,换成自己的密钥或路径。

  3. 3
    写入客户端配置

    在客户端的 MCP 配置里填好启动命令和参数,保存后完全退出客户端再打开,让新配置生效。

  4. 4
    验证是否可用

    连接成功后客户端会列出约 3 个工具,随便调一个只读工具确认能返回结果。

典型使用场景

3 类
按分类与标签整理的常见用法,具体以官方文档为准
文档检索问答

在你的文档库里按问题找答案。

知识库读写

把新资料写进知识库,随用随取。

内容整理

把零散资料归纳成结构化笔记。

选择建议

选之前先看这几条
按条目字段给出的判断,不替你决定用不用

值得选的地方

  • 资料标注 3 个工具,职责相对集中,接入后不容易误操作。
  • 有现成的安装命令,不用自己找入口,照着跑就能起服务。
  • 已适配 Claude Desktop、Claude,换客户端不用重新研究配置。
  • 资料标注无需鉴权,装好就能连。

先想清楚的地方

  • stdio 方式要求本地装好运行时(如 Node.js),换机器要重新装一遍。
  • 版本更新与行为细节以官方仓库为准。

更适合

  • 需要在文档知识场景里让模型直接调用外部能力的人。
  • 已经在用 Claude Desktop、Claude 的用户,接入路径最短。
  • 想先快速试一下、能接受在本地或自有环境跑服务的人。

不太适合

  • 完全不想在本地装环境的人。
  • 需要厂商级 SLA 和技术支持的生产环境。

同类对比

6 个同类
同分类里挑最接近的几条,字段以各自详情页为准
MCP 服务器工具数传输方式部署方式安装命令最后核验
Obsidian Rag 1a1fd3d3当前 3 stdio · 本地进程 本地部署 有 2026-05-17
dify 未标注 stdio · 本地进程 本地部署 有 2026-04-11
open-webui 未标注 stdio · 本地进程 本地部署 有 2026-04-11
JeecgBoot 未标注 stdio · 本地进程 本地部署 有 2026-04-11
Cherry Studio 未标注 未标注 未标注 无 2026-05-17
mempalace 未标注 stdio · 本地进程 本地部署 有 2026-04-11

服务器档案

文档知识
字段来自项目仓库与公开索引,中文说明由本站标注
提供方GabrielFersPin上游标注的维护方 GabrielFersPin
分类文档知识本站按上游分类展示
连接方式stdio · 本地进程客户端在本机拉起子进程,数据不出本机
鉴权方式无需鉴权远程服务建议用最小权限凭据
维护方社区 / 第三方核验等级 仅资料核对
包名 / 运行时未标注 · Python本地安装时用到的包标识与运行时
部署类型本地部署上游给出的部署形态
最近核验2026-05-17上游最后核验时间

详细介绍

23.6 KB
来自项目仓库与上游目录的原文,未做删改

🧠 Obsidian RAG System - User Manual

> RAG system to transform your Obsidian vault into an intelligent study assistant

📋 Table of Contents

  • What is this?
  • Features
  • Requirements
  • Installation
  • Configuration
  • MCP Integration
  • Usage
  • Use Cases
  • Troubleshooting
  • Architecture

🎯 What is this?

A Retrieval-Augmented Generation (RAG) system that allows you to query your personal Obsidian knowledge base using artificial intelligence.

What problems does it solve?

If you take notes in Obsidian, you probably face:
- ❌ Manually searching through hundreds of notes is slow
- ❌ You forget connections between concepts from different topics
- ❌ You don't leverage all the knowledge you've documented

With this RAG system:
- ✅ Instant semantic search across all your notes
- ✅ Find relevant information even when using different words
- ✅ Automatically discover connections between topics
- ✅ Everything runs locally on your PC (privacy guaranteed)


✨ Features

  • 🔍 Intelligent semantic search: Find notes by meaning, not just exact word matches
  • 📊 Subject/topic filtering: Organize your searches by categories (Obsidian tags)
  • 💾 Local vector database: Your data never leaves your computer
  • 🚀 Fast and efficient: Optimized embedding model (384 dimensions)
  • 📈 Detailed statistics: See how many notes you have indexed per topic
  • 🔄 Incremental updates: Reindex only when you add new notes
  • 💬 Interactive mode: Conversational interface for multiple queries
  • 🔌 MCP Integration: Use your vault directly from Claude Desktop and Claude Code

💻 Requirements

Minimum system requirements

  • Operating System: Linux, macOS, or Windows with WSL2
  • Python: 3.10 or higher
  • RAM: Minimum 4GB (8GB recommended)
  • Disk space: ~2GB for models and database

Required software

  • Python 3.10+
  • pip (Python package manager)
  • git (to clone the repository)
  • An Obsidian vault with notes in Markdown format

🚀 Installation

Step 1: Clone the repository```bash

git clone https://github.com/GabrielFersPin/obsidian-rag-system.git
cd obsidian-rag-system


### Step 2: Create virtual environment```bash
# Create virtual environment
python3 -m venv venv

# Activate virtual environment
# On Linux/macOS:
source venv/bin/activate

# On Windows (WSL2):
source venv/bin/activate

Step 3: Install dependencies```bash

pip install -r requirements.txt


**Note**: The first installation will download ~1GB of dependencies, including PyTorch and Sentence Transformers.

---

## ⚙️ Configuration

### Step 1: Configure environment variables

Create a `.env` file in the project root:```bash
cp .env.example .env

Edit the .env file with your configuration:```env

Path to your Obsidian vault

OBSIDIAN_VAULT_PATH=/home/your-user/Documents/MyVault

Embedding model (you can change it if desired)

EMBEDDING_MODEL=sentence-transformers/all-MiniLM-L6-v2

Directory for the vector database

CHROMA_PERSIST_DIR=./chroma_db


### Step 2: Create symbolic link to vault (Optional)

If you prefer not to specify the full path, you can create a symbolic link:```bash
# Create the data directory if it doesn't exist
mkdir -p data

# Create the symbolic link to your vault
ln -s /path/to/your/obsidian/vault ./data/vault

Then in .env:```env
OBSIDIAN_VAULT_PATH=./data/vault


### Step 3: Recommended note structure

The system works best if your notes have:

**YAML metadata (optional but recommended):**```yaml
---
deck: DataScience
tags: [machine-learning, python]
created: 2024-01-15
---

The deck field is used to filter searches by subject/topic.


🔌 MCP Integration (Model Context Protocol)

This system includes an MCP server that allows Claude Desktop and Claude Code to search your vault directly from their interface.

What is MCP?

MCP (Model Context Protocol) allows AI assistants like Claude to access external tools and data sources. With this integration, Claude can:
- 🔍 Search your Obsidian vault in real-time
- 📊 Get vault statistics
- 📚 List notes by subject
- 💬 Answer questions using your personal knowledge base

Available MCP Tools

  1. search_vault: Search for information in your vault
    - Parameters: query (required), deck (optional filter), n_results (1-10)

  2. get_vault_stats: Get vault statistics
    - Shows total notes, distribution by subject, last indexing time

  3. list_notes_by_deck: List all notes from a specific subject
    - Parameters: deck (Algorithms, Cloud, DataScience, Architecture)

Setup for Claude Desktop

Claude Desktop allows you to connect local MCP servers. Here's how:

Step 1: Locate Claude Desktop config file

The configuration file location depends on your OS:

macOS:```bash
~/Library/Application Support/Claude/claude_desktop_config.json


**Windows:**```bash
%APPDATA%\Claude\claude_desktop_config.json

Linux:```bash
~/.config/Claude/claude_desktop_config.json


#### Step 2: Edit the config file

Open the config file and add your MCP server configuration:```json
{
  "mcpServers": {
    "obsidian-rag": {
      "command": "/FULL/PATH/TO/obsidian-rag-system/venv/bin/python",
      "args": [
        "/FULL/PATH/TO/obsidian-rag-system/mcp_server.py"
      ],
      "env": {
        "OBSIDIAN_VAULT_PATH": "/FULL/PATH/TO/YOUR/VAULT"
      }
    }
  }
}

Important: Replace the paths with your actual paths:
- /FULL/PATH/TO/obsidian-rag-system/ → Your project directory
- /FULL/PATH/TO/YOUR/VAULT → Your Obsidian vault directory

Example (macOS/Linux):```json
{
"mcpServers": {
"obsidian-rag": {
"command": "/FULL/PATH/TO/obsidian-rag-system/venv/bin/python",
"args": [
"/FULL/PATH/TO/obsidian-rag-system/mcp_server.py"
],
"env": {
"OBSIDIAN_VAULT_PATH": "/FULL/PATH/TO/VAULT"
}
}
}
}


**Example (Windows):**```json
{
  "mcpServers": {
    "obsidian-rag": {
      "command": "C:\\Users\\YourName\\obsidian-rag-system\\venv\\Scripts\\python.exe",
      "args": [
        "C:\\Users\\YourName\\obsidian-rag-system\\mcp_server.py"
      ],
      "env": {
        "OBSIDIAN_VAULT_PATH": "C:\\Users\\YourName\\Documents\\MyVault"
      }
    }
  }
}

Step 3: Restart Claude Desktop

Close and reopen Claude Desktop completely.

Step 4: Verify the connection

In Claude Desktop, you should see a 🔌 icon or tools indicator. You can now ask:```
"Search my vault for information about QuickSort"
"What are my notes on Docker?"
"Show me statistics of my vault"


### Setup for Claude Code (CLI)

Claude Code can also use MCP servers. The setup is different:

#### Step 1: Create MCP config file

Create or edit the MCP configuration file for Claude Code:

**Location:** `~/.config/claude-code/mcp.json` (Linux/macOS) or `%APPDATA%\claude-code\mcp.json` (Windows)```bash
# Create directory if it doesn't exist
mkdir -p ~/.config/claude-code

# Create or edit the config file
nano ~/.config/claude-code/mcp.json

Step 2: Add server configuration```json

{
"mcpServers": {
"obsidian-rag": {
"command": "/FULL/PATH/TO/obsidian-rag-system/venv/bin/python",
"args": [
"/FULL/PATH/TO/obsidian-rag-system/mcp_server.py"
],
"env": {
"OBSIDIAN_VAULT_PATH": "/FULL/PATH/TO/YOUR/VAULT",
"CHROMA_PERSIST_DIR": "/FULL/PATH/TO/obsidian-rag-system/chroma_db"
}
}
}
}


**Example (Linux/macOS):**```json
{
  "mcpServers": {
    "obsidian-rag": {
      "command": "/home/username/obsidian-rag-system/venv/bin/python",
      "args": [
        "/home/username/obsidian-rag-system/mcp_server.py"
      ],
      "env": {
        "OBSIDIAN_VAULT_PATH": "/home/username/Documents/MyVault",
        "CHROMA_PERSIST_DIR": "/home/username/obsidian-rag-system/chroma_db"
      }
    }
  }
}

Example (Windows):```json
{
"mcpServers": {
"obsidian-rag": {
"command": "C:\Users\YourName\obsidian-rag-system\venv\Scripts\python.exe",
"args": [
"C:\Users\YourName\obsidian-rag-system\mcp_server.py"
],
"env": {
"OBSIDIAN_VAULT_PATH": "C:\Users\YourName\Documents\MyVault",
"CHROMA_PERSIST_DIR": "C:\Users\YourName\obsidian-rag-system\chroma_db"
}
}
}
}


#### Step 3: Test the server

You can test if the MCP server works correctly:```bash
# Activate virtual environment
source /path/to/obsidian-rag-system/venv/bin/activate

# Run the MCP server directly (for testing)
python /path/to/obsidian-rag-system/mcp_server.py

If everything is configured correctly, the server should start without errors (it will wait for input via stdio).

Step 4: Use in Claude Code

When you start Claude Code, it will automatically load the MCP server. You can use it by asking Claude to search your vault:```
"Use the obsidian-rag server to search my notes about algorithms"
"Get statistics from my vault using the MCP tool"


### Troubleshooting MCP Setup

#### Server not appearing in Claude Desktop

**Problem**: The 🔌 icon doesn't appear or tools aren't available.

**Solutions**:
1. Verify the JSON syntax is correct (no trailing commas, proper quotes)
2. Check that all paths are absolute (not relative like `./` or `~/`)
3. Verify the Python executable exists: `ls /path/to/venv/bin/python`
4. Check the vault path exists: `ls /path/to/vault`
5. Restart Claude Desktop completely (quit and reopen)

#### Error: "ModuleNotFoundError"

**Problem**: Python can't find the required modules.

**Solution**: Make sure you're using the Python from the virtual environment:```bash
/FULL/PATH/TO/obsidian-rag-system/venv/bin/python

Not the system Python (/usr/bin/python3).

Error: "Vault not indexed"

Problem: The vault hasn't been indexed yet.

Solution: Run the indexing first:```bash
cd /path/to/obsidian-rag-system
source venv/bin/activate
python run_rag.py

Select option 2 to index


#### MCP server starts but searches fail

**Problem**: ChromaDB can't find the database.

**Solution**: Add the `CHROMA_PERSIST_DIR` environment variable to your MCP config:```json
"env": {
  "OBSIDIAN_VAULT_PATH": "/path/to/vault",
  "CHROMA_PERSIST_DIR": "/path/to/obsidian-rag-system/chroma_db"
}

Testing the MCP server manually

You can test the server manually with a simple script:```bash
cd /path/to/obsidian-rag-system
source venv/bin/activate

Run server in test mode

python -c "
import asyncio
from mcp_server import get_rag_system

async def test():
rag = get_rag_system()
info = rag.get_system_info()
print(f'Notes indexed: {info[\"indexed_notes\"]}')

asyncio.run(test())
"


### Example Usage with Claude

Once configured, you can interact with your vault naturally:

**Example 1: Search for specific topic**```
You: "Search my vault for notes about Docker containers"

Claude: *Uses search_vault tool*
Found 3 relevant notes:
1. Docker Introduction.md - Explains containerization...
2. Containers vs VMs.md - Compares containers with VMs...

Example 2: Get vault overview```
You: "Show me statistics of my note collection"

Claude: Uses get_vault_stats tool
Your vault has 98 indexed notes:
- general: 84 notes
- DataScience: 3 notes
- Algorithms: 2 notes
...


**Example 3: Filter by subject**```
You: "What notes do I have about algorithms?"

Claude: *Uses search_vault with deck filter*
Found notes in Algorithms:
- QuickSort.md
- Binary Search.md
- Sorting Algorithms.md

📖 Usage

Start the system```bash

Make sure the virtual environment is activated

source venv/bin/activate

Run the main program

python3 run_rag.py


### Main menu

When you start, you'll see a menu with the following options:```
============================================================
MAIN MENU
============================================================
1. 📊 View system statistics
2. 🔄 Index vault (first time or update)
3. 🔍 Interactive search
4. 💬 Single query
5. 🧪 Query examples
6. 🚪 Exit
============================================================

First time: Index your vault

IMPORTANT: Before searching, you must index your notes.

  1. Select option 2 (Index vault)
  2. Confirm that you want to proceed (type s and press Enter)
  3. Wait while the system processes all your notes
  4. Indexing can take 1-10 minutes depending on your vault size

Indexing process:```
⏳ Starting indexing...

📚 Step 1/5: Loading notes from vault...
✅ 103 notes loaded

🔍 Step 2/5: Filtering valid notes...
✅ 98 valid notes

📝 Step 3/5: Creating collection 'obsidian_notes'...
✅ Collection created

🧮 Step 4/5: Generating embeddings...
⏳ Processing batch 1/2...
⏳ Processing batch 2/2...
✅ Embeddings generated

💾 Step 5/5: Storing in ChromaDB...
✅ Documents stored

✅ Indexing completed!
📚 Notes processed: 98
⏱️ Total time: 45.32s


### View statistics (Option 1)

Shows information about your database:```
📊 SYSTEM STATISTICS
============================================================
📁 Vault: /home/user/vault
📚 Indexed notes: 98
🕐 Last indexing: 2025-11-18T13:52:45
🧮 Model: sentence-transformers/all-MiniLM-L6-v2
📏 Dimension: 384

🎴 Distribution by subject:
   general                         84 notes
   DataScience                      3 notes
   Algorithms                       2 notes
   Cloud                            5 notes
   Architecture                     2 notes
============================================================

Interactive search (Option 3)

Conversational mode for multiple queries:```
🔍 INTERACTIVE SEARCH MODE
============================================================
Type 'exit' to return to the main menu

💭 Your question: What is QuickSort?
🎴 Filter by subject (Enter for all): Algorithms

🔍 Searching in 2 notes from Algorithms...

📝 RETRIEVED CONTEXT:

QuickSort is an efficient sorting algorithm that uses
the divide and conquer paradigm...

📚 SOURCES:
1. [Algorithms] QuickSort.md (Relevance: 95%)
2. [Algorithms] Sorting Algorithms.md (Relevance: 78%)
============================================================

💭 Your question: exit


### Single query (Option 4)

For a quick single query:```
💬 SINGLE QUERY
============================================================
💭 Your question: Explain Docker
🎴 Filter by subject (Enter for all): [Enter]

🔍 Searching...

📝 RETRIEVED CONTEXT:
============================================================
Docker is a container platform that allows packaging
applications with all their dependencies...

📚 SOURCES:
   1. [Cloud] Docker Introduction.md (Relevance: 92%)
   2. [Cloud] Containers vs VMs.md (Relevance: 85%)
============================================================

Predefined examples (Option 5)

Try example queries to familiarize yourself with the system.


🎓 Use Cases

1. Study for an exam```

💭 Your question: Summarize key concepts of neural networks
🎴 Filter by subject: DataScience


The system will find all your notes related to neural networks and show you a consolidated summary.

### 2. Connect concepts between subjects```
💭 Your question: How does CPU pipeline relate to Docker?
🎴 Filter by subject: [Enter - search all]

Discover connections you hadn't noticed between different study areas.

3. Remember something you wrote months ago```

💭 Your question: That note where I talked about algorithmic complexity and examples with trees
🎴 Filter by subject: [Enter]


Semantic search helps you find notes even when you don't remember the exact words.

### 4. Prepare presentations```
💭 Your question: Give me all the important concepts about microservices
🎴 Filter by subject: Cloud

Gather information from multiple notes to create complete presentations.


🔧 Troubleshooting

Error: "ModuleNotFoundError: No module named 'dotenv'"

Problem: Virtual environment not activated.

Solution:```bash
source venv/bin/activate
pip install -r requirements.txt


### Error: "Vault not found at [path]"

**Problem**: The vault path in `.env` is incorrect.

**Solution**:
1. Verify that the `.env` file exists
2. Verify that `OBSIDIAN_VAULT_PATH` points to the correct folder
3. Verify that the folder exists: `ls /path/to/your/vault`

### System doesn't find notes with certain words

**Problem**: The search is semantic, not literal.

**Explanation**: The system searches by meaning, not exact word matching. If you search for "ML" it might not find notes that say "Machine Learning" unless there's sufficient context.

**Solution**: Use more descriptive phrases instead of abbreviations.

### Indexing takes too long

**Problem**: You have many notes or a slow CPU.

**Solution**:
- It's normal on first indexing (1-10 min)
- The model uses CPU by default for compatibility
- Subsequent indexings will be faster

### "No notes indexed"

**Problem**: You haven't run the indexing.

**Solution**:
1. Go to the main menu
2. Select option `2` (Index vault)
3. Confirm with `s`

### Errors with special characters

**Problem**: Some notes have non-UTF-8 characters.

**Solution**: Make sure your Markdown notes are in UTF-8. In Obsidian this is the standard.

---

## 🏗️ Architecture

### How does it work internally?```
┌─────────────────────────────────────────────┐
│           OBSIDIAN RAG SYSTEM               │
├─────────────────────────────────────────────┤
│                                             │
│  📝 YOUR OBSIDIAN VAULT                     │
│      └─ Markdown notes                      │
│      └─ YAML metadata (deck, tags)          │
│           ↓                                 │
│  🔍 OBSIDIAN LOADER                         │
│      └─ Reads .md files                     │
│      └─ Extracts metadata                   │
│      └─ Ignores templates/.trash            │
│           ↓                                 │
│  🧮 SENTENCE TRANSFORMERS                   │
│      └─ Converts text to vectors            │
│      └─ Model: all-MiniLM-L6-v2             │
│      └─ Dimension: 384                      │
│           ↓                                 │
│  💾 CHROMADB (Vector Database)              │
│      └─ Stores vectors + metadata           │
│      └─ Similarity search (cosine)          │
│      └─ Local disk persistence              │
│           ↓                                 │
│  🔎 SEARCH SYSTEM                           │
│      └─ Top-K most relevant documents       │
│      └─ Filters by deck/subject             │
│      └─ Ranking by relevance                │
│           ↓                                 │
│  💬 USER INTERFACE                          │
│      └─ Interactive menu                    │
│      └─ Natural language queries            │
│                                             │
└─────────────────────────────────────────────┘

Main components

Component Function Technology
Loader Reads and parses Markdown notes Python + YAML
Embeddings Converts text to vectors Sentence Transformers
Vector DB Stores and searches vectors ChromaDB
RAG Chain Orchestrates the entire pipeline Python
Interface User menu Interactive CLI

Query flow

  1. User asks question: "What is QuickSort?"
  2. Question → Vector: The question is converted to a 384-dimension vector
  3. Semantic search: ChromaDB searches for the most similar vectors
  4. Ranking: Orders results by relevance (cosine similarity)
  5. Response: Shows the most relevant notes and context

📁 Project Structure```

obsidian-rag-system/
├── src/ # Source code
│ ├── obsidian_loader.py # Reads Obsidian notes
│ ├── embeddings.py # Generates vectors
│ ├── vectorstore.py # Manages ChromaDB
│ └── rag_chain.py # Complete pipeline
│
├── run_rag.py # Main script (run this)
├── requirements.txt # Python dependencies
├── .env # Configuration (create this)
├── .env.example # Configuration template
│
├── venv/ # Virtual environment (create with python -m venv venv)
├── chroma_db/ # Vector database (created automatically)
└── data/ # Link to your vault (optional)
└── vault/ # Symlink to Obsidian


---

## 🔄 System Updates

### When you add new notes```bash
# 1. Run the program
python3 run_rag.py

# 2. Select option 2 (Index vault)

# 3. Confirm reindexing

Note: The system replaces the complete index. This ensures there are no duplicates.

Update dependencies```bash

source venv/bin/activate
pip install --upgrade -r requirements.txt


---

## 🛡️ Privacy and Security

- ✅ **Everything is local**: Your notes never leave your computer
- ✅ **No telemetry**: ChromaDB may report anonymous statistics, but your data is not shared
- ✅ **No API keys**: No external service API keys required
- ✅ **Open source**: You can audit all the code

---

## 💡 Tips and Best Practices

### 1. Organize your notes with metadata

Add YAML frontmatter to your notes:```markdown
---
deck: DataScience
tags: [python, jupyter, pandas]
created: 2025-01-15
---

# Introduction to Pandas

Pandas is a library...

2. Use the deck field to categorize

The deck field is used to filter searches:

  • deck: DataScience
  • deck: Algorithms
  • deck: Cloud

3. Write complete questions

❌ Bad: "quicksort"
✅ Good: "How does the QuickSort algorithm work?"

Semantic search works better with context.

4. Reindex periodically

Recommendation: Reindex once a week if you add many notes.

5. Clean empty notes or templates

The system automatically ignores:
- Folders named template or .trash
- Very short notes (< 50 characters)
- .obsidian and .git files


📊 Technology Stack

Technology Version Purpose
Python 3.10+ Main language
Sentence Transformers 2.2.0+ Embedding generation
ChromaDB 0.4.22+ Vector database
PyTorch 2.0.0+ ML backend
python-dotenv 1.0.0+ Environment variables

🤝 Contributing

This project is open to contributions:

  1. Fork the repository
  2. Create a branch for your feature (git checkout -b feature/new-functionality)
  3. Commit your changes (git commit -m &#x27;Add new functionality&#x27;)
  4. Push to the branch (git push origin feature/new-functionality)
  5. Open a Pull Request

📄 License

MIT License - You are free to use, modify, and distribute this software.

See LICENSE for more details.


📧 Support

If you have problems:

  1. Review the Troubleshooting section
  2. Search in Issues
  3. Open a new Issue with:
    - Problem description
    - Steps to reproduce
    - Error logs (if any)
    - Your operating system and Python version

🙏 Acknowledgments

  • Sentence Transformers - Embedding models
  • ChromaDB - Vector database
  • Obsidian - Note-taking system
  • Niklas Luhmann's Zettelkasten method

Last updated: November 2025


> 💡 Tip: If you find this project useful, give it a ⭐ on GitHub and share it with other students who use Obsidian.

常见问题

8 问
答案由本站依据条目字段整理,不引用上游原文
Obsidian Rag 1a1fd3d3 怎么安装?

在客户端按下面的安装命令接入:python3 -m venv venv。装完在客户端配置里引用同一个包或端点。

Obsidian Rag 1a1fd3d3 怎么在 Claude Desktop、Claude 里配置?

它的传输方式是stdio · 本地进程,在客户端的 MCP 配置里填好启动命令和参数,保存后完全退出客户端再打开,配置才会生效。

Obsidian Rag 1a1fd3d3 收费吗?

本站不代收费用,也没有标注具体价格。是否收费取决于服务本身和背后的接口额度,正式使用前按官方说明确认。

Obsidian Rag 1a1fd3d3 需要 API Key 吗?

资料标注为无需认证,装好就能连;如果实际调用时报鉴权错误,按官方文档补上凭据。

Obsidian Rag 1a1fd3d3 支持哪些客户端?

资料标注支持 Claude Desktop、Claude;其他客户端只要能配置 MCP Server,也能用同样的命令或地址接入。

Obsidian Rag 1a1fd3d3 能提供哪些能力?

资料显示约 3 个工具;能力标签包括 知识管理、本地AI、Python、Claude。具体工具清单以官方文档为准。

Obsidian Rag 1a1fd3d3 需要什么运行环境?

运行时为 Python;属于本地进程,需要在本机装好运行环境。

Obsidian Rag 1a1fd3d3 是本地还是远程?

它是本地进程(stdio),由客户端在本机拉起。传输方式为stdio · 本地进程。

热门专题

按标签挑服务器
全部专题 →

服务器信息以作者仓库与官方文档为准;分类、传输方式与鉴权方式的中文说明为中文整理,最近核验 2026-05-17。