広告掲載
MCP.so
ログイン

Obsidian MCP Server

@roelven

リポジトリ

基本情報

カテゴリ

メモリとナレッジ

ライセンス

MIT license

トランスポート

stdio

作者

roelven

Obsidian MCP Server について

MCP (Model Context Protocol) server for Obsidian notes access via CouchDB LiveSync

#obsidian#mcp-server#couchdb

Obsidian MCP Server の設定

以下の設定を使って、このサーバーを MCP 対応クライアントに追加してください。

{
  "mcpServers": {
    "obsidian-mcp-server-roelven": {
      "command": "docker",
      "args": [
        "build",
        "-t",
        "obsidian-mcp-server",
        "."
      ]
    }
  }
}

Obsidian MCP Server のツール

ツール一覧はまだありません

この掲載には表示できるツール一覧がまだありません。サーバーにツールがないという意味ではありません。対応機能はプロジェクトのドキュメントをご確認ください。

ツールは README から自動的に抽出されます。メンテナーは ## Tools という見出しの下に記載することで、このタブに反映できます。

概要

What is Obsidian MCP Server?

A Model Context Protocol (MCP) server that gives AI models read-only access to your Obsidian notes by connecting through your existing LiveSync CouchDB setup. It is intended for users who want to query, search, and browse their Obsidian vault programmatically via MCP clients like Claude Desktop.

How to use Obsidian MCP Server?

Install via Docker (recommended) or run locally with Python 3.10+. Configure required environment variables (COUCHDB_BASE_URL, COUCHDB_DATABASE_NAME, COUCHDB_USER, COUCHDB_PASSWORD, API_KEY) in a .env file. Start the server with obsidian-mcp-server --transport stdio for direct MCP client connections or --transport sse --port 8000 for HTTP-based access. Integrate with clients like Claude Desktop by adding a mcpServers entry that runs the Docker image with the .env file.

Key features of Obsidian MCP Server

  • Read-only access to Obsidian notes via LiveSync CouchDB
  • Performance-optimized resource listing (10 recent notes)
  • Automatic content inclusion for small result sets (≤3 notes)
  • Metadata extraction: frontmatter, tags, and aliases
  • Reassembles chunked notes automatically
  • Supports encrypted vaults (if VAULT_PASSPHRASE is set)
  • Docker support and full environment variable configuration

Use cases of Obsidian MCP Server

  • AI assistants answering questions by searching your Obsidian notes
  • Browsing recent notes sorted by modification, creation time, or path
  • Quickly retrieving the most recent note with full content
  • Integrating with Claude Desktop to query your vault without direct file access
  • Searching notes by title, content, or tags from any MCP‑compatible client

よくある質問

1

What prerequisites are needed?

A running Obsidian LiveSync CouchDB instance with read‑access credentials, plus Python 3.10+ (local) or Docker.

2

Can Obsidian MCP Server read encrypted notes?

Yes. Set the `VAULT_PASSPHRASE` environment variable to decrypt encrypted Obsidian LiveSync notes. If not set, encrypted notes will not be decrypted.

3

Does Obsidian MCP Server support write operations?

No. The server provides read‑only access to your notes for safety. Write operations are listed as a future enhancement.

4

What transport protocols are available?

The server supports two MCP transports: `stdio` (direct client connection) and `SSE` (HTTP‑based on a configurable port, default 8000).

5

How can I improve query performance?

Create a JSON index on the `type` and `mtime` fields in your LiveSync CouchDB database. This index is recommended for efficient listing and sorting of notes.

コメント

「メモリとナレッジ」の他のコンテンツ

Stele

Stele-Dev

Shared memory for AI coding agents. Claude Code, Cursor and Codex read your project's decisions, lessons and tasks before they act.

Neither MCP

stonianua

Project context your AI can query through MCP — with source evidence.

priostack

ideaswave

Open-source Python SDK & MCP interface for Priostack Agent Context Network (ACN). Zero-setup, autonomous memory orchestration for AI agents.

Sensefold

SenseFold Labs

Personal context for every AI: search, read, and write back to your private Markdown library of articles, threads, PDFs, notes, and captured ChatGPT/Claude/Gemini/Grok conversations. Remote server, OAuth 2.1 paste-and-au

Azquo

Azquo Holdings Ltd

Exact, shared memory for AI agents: a hierarchical database where totals always add up, every write is attributed to the agent and person behind it, and any figure traces back to its source file. Agents create databases,

広告掲載