ZMCP
ZMCP turns your Obsidian vault into an AI-accessible workspace. It creates a bridge between your notes and AI assistants like Claude, allowing them to read, write, and manage your knowledge base directly.
README
ZMCP - Obsidian MCP Server
What it does
ZMCP turns your Obsidian vault into an AI-accessible workspace. It creates a bridge between your notes and AI assistants like Claude, allowing them to read, write, and manage your knowledge base directly.
Think of it as giving AI tools permission to work with your notes - they can search through your documentation, create new notes, update existing ones, and help manage tasks across your entire vault.
Tools & Capabilities
Note Operations
- Browse & Search: AI can find and read any note in your vault instantly
- Create & Edit: Generate new notes or update existing ones based on your requests
- Organize: Manage tags, find related content, and maintain consistency across notes
<details> <summary><h3>Task Management</h3></summary>
- Smart Task Queries: Find tasks by status, priority, due dates, or tags
- Create with Dates: Add tasks with due dates, scheduled dates, and recurrence patterns
- Status Updates: Mark tasks complete, change priorities, or update any task properties
- Context-Aware: AI understands where tasks belong in your notes and maintains your formatting
</details>
<details> <summary><h3>Flashcard Learning System</h3></summary>
Transform any learning material into structured flashcard decks. Ask AI to research topics, extract key concepts, and automatically create study materials in your vault. Works with Spaced repetition plugin.
Example prompt: "Research OAuth 2.0 authentication flow and create a flashcard deck with the core concepts, grant types, and security best practices"
Result: AI fetches documentation, extracts essential knowledge, and creates 20+ flashcards like:
- "What are the four OAuth 2.0 grant types?" → "Authorization Code, Implicit, Client Credentials, Resource Owner Password"
- "The ==authorization code== grant type is recommended for server-side applications"
</details>
Plugin Intelligence
- Vault Awareness: AI can see what Obsidian plugins you have installed
- Workflow Integration: Works alongside your existing Obsidian setup without disruption
Prompts & Resources
The real power comes from turning your notes into reusable AI instructions:
Dynamic Prompts & Commands
Configure which folders or tags the server monitors for prompts - any note in those locations becomes an AI instruction template. Access them instantly with / in Claude or Cursor. Edit the note in Obsidian, and AI assistants see the changes immediately. Create templates with {{placeholders}} that AI fills dynamically, perfect for standardized workflows like bug reports, meeting notes, or code reviews.
Resources
Keep coding standards, writing guidelines, or project requirements in your vault. Reference them with @ in Claude Code or Cursor to give AI instant context. AI assistants will follow these rules consistently across all interactions. No more copy-pasting instructions or explaining context repeatedly.
Real-time Sync
Changes to prompts and resources sync instantly. Edit your guidelines while AI is working, and it adapts on the fly. Your knowledge base becomes a living instruction set that evolves with your projects.
Installation
Plugin Installation
Install using BRAT (Beta Reviewers Auto-update Tool), which lets you install Obsidian plugins directly from GitHub:
- Search for "BRAT" in Obsidian's Community Plugins and install it
- In BRAT settings, click "Add Beta Plugin"
- Enter:
https://github.com/lrangell/zmcp - Enable ZMCP in your Community Plugins list
- Configure your prompt folders and tags in the plugin settings
Client Configuration
Claude Code
claude mcp add zmcp -t http http://localhost:3983/mcp
Claude Desktop
Add to your claude_desktop_config.json:
{
"mcpServers": {
"zmcp": {
"command": "npx",
"args": ["mcp-remote", "http://localhost:3983/mcp"]
}
}
}
Cursor
Or manually add to Cursor settings:
{
"mcp": {
"servers": {
"zmcp": {
"transport": "http",
"url": "http://localhost:3983/mcp"
}
}
}
}
Implementation
Built on top of ts-mcp-forge, a TypeScript framework for creating MCP servers with decorators. The plugin runs an HTTP server inside Obsidian's Electron environment, exposing vault operations as MCP tools and notes as resources. Uses Obsidian's Vault API for all file operations, ensuring compatibility with sync and other plugins.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。