Provides read-only access to Obsidian vaults with semantic search, tag-based filtering, path-based filtering, temporal queries, metadata filtering, note retrieval, and smart summarization capabilities
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Obsidian MCP Second Brain Serverfind notes about project planning from the last month"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Obsidian MCP Second Brain Server
A read-only MCP server for intelligent, secure access to your Obsidian vault—enabling semantic search, metadata filtering, and more for LLMs.
Features
Efficient Database Storage: SQLite-based indexing for large vaults with persistent caching
Memory Mode Option: Optional in-memory indexing for small vaults or development
Semantic Search: Full-text search across all notes with fuzzy matching
Tag-Based Filtering: Search by hierarchical tags (e.g.,
work/puppet,tech/golang)Path-Based Filtering: Filter by directory patterns (e.g.,
Work/Puppet/**)Temporal Queries: Filter notes by creation/modification dates
Metadata Filtering: Filter by type, status, and category
Note Retrieval: Get full content of specific notes
Smart Summarization: Generate summaries of note collections
Recent Notes: Quick access to recently modified notes
Archive Control: Optionally include archived notes in searches
Security: Path traversal protection, file size limits, input validation
Read-Only Design
This MCP server is intentionally read-only to ensure your vault remains safe during AI interactions. It provides:
✅ Search and retrieve notes
✅ Filter by metadata and paths
✅ Generate summaries and statistics
❌ No note creation or editing
❌ No file modifications
For write operations, consider using dedicated Obsidian plugins with built-in safety checks.
Installation
Configuration & Installation
One-Click Installation
VS Code:
VS Code Insiders:
Cursor:
Manual Installation
No installation needed! Use directly with npx:
Local Development
This makes the server available globally as obsidian-mcp-sb.
Claude Code & Claude Desktop
Claude Code
Add the server using:
Claude Desktop
Add to your Claude Desktop configuration file:
macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%/Claude/claude_desktop_config.json
See docs/configuration.md for vault structure, CLI arguments, and configuration examples.
See docs/api.md for the full MCP API reference and usage examples.
Usage with Claude Code
Add to your MCP configuration file.
Single Vault Configuration
macOS/Linux: Edit ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: Edit %APPDATA%\Claude\claude_desktop_config.json
Multiple Vault Configuration
You can configure multiple vault instances:
Local Development Setup
If you're developing locally with npm link:
See docs/examples.md for more example queries.
Documentation
See docs/README.md for:
API Reference
Configuration & CLI Options
Example Queries
Development & Storage Architecture
Contributing
Dependencies
Architecture & Database Schema
Security & Search
Example Queries
Development
Contributing
Dependencies
Quick Start
Run the server instantly with npx (no install required):
Or add to Claude Code/Claude Desktop (see Configuration & Installation below).
Troubleshooting & FAQ
See docs/configuration.md for common issues and solutions.
See docs/search.md for details on search weights and scoring.
See docs/security.md for details on security features and protections.
See docs/development.md for development workflow and storage details.
See docs/contributing.md for contribution guidelines.
Storage Architecture
The server uses SQLite database storage by default for efficient indexing and persistent caching:
Database Mode (Default): Stores indexed notes in
.obsidian-mcp/notes.dbwithin your vaultPersistent indexing (survives server restarts)
Efficient for large vaults (1000+ notes)
Full-text search with SQLite FTS5
Lower memory usage
Memory Mode (Optional): Use
--use-memoryflag for in-memory storageFaster for small vaults (<100 notes)
No disk I/O overhead
Useful for development and testing
Uses Fuse.js for fuzzy search
See docs/architecture.md for the architecture diagram. See docs/database-schema.md for the database schema.
Architecture
src/index.ts: MCP server implementation with tool handlerssrc/vault.ts: Vault indexing orchestration and security controlssrc/storage.ts: Storage interface abstractionsrc/database-storage.ts: SQLite-based storage implementationsrc/memory-storage.ts: In-memory storage implementation with Fuse.jssrc/storage-factory.ts: Storage factory pattern for mode selectionsrc/config.ts: Configuration management with defaultssrc/types.ts: TypeScript type definitions and validation utilitiessrc/__tests__/: Unit tests for critical functionality
See docs/dependencies.md for a full list of production and development dependencies.
Support
If obsidian-mcp-sb saves you time, consider sponsoring me on GitHub or buy me a coffee.
License
MIT