mcp-obsidian-second-brain
MCP server that provides persistent, structured memory using an Obsidian vault with the PARA method.
README
mcp-obsidian-second-brain
An MCP (Model Context Protocol) server that gives Claude a persistent, structured memory system backed by an Obsidian vault. Memories are stored as Markdown files with YAML frontmatter using the PARA method (Projects, Areas, Resources, Archives).
Installation
Option 1: npx from GitHub (no clone required)
Add to your Claude Desktop config (~/Library/Application Support/Claude/claude_desktop_config.json on macOS):
{
"mcpServers": {
"obsidian-second-brain": {
"command": "npx",
"args": ["-y", "@neverprepared/mcp-obsidian-second-brain"],
"env": {
"OBSIDIAN_VAULT_PATH": "/path/to/your/obsidian/vault"
}
}
}
}
Option 2: Clone and run locally
git clone https://github.com/neverprepared/mcp-obsidian-second-brain.git
cd mcp-obsidian-second-brain
npm install
npm run build
Then add to your Claude Desktop config:
{
"mcpServers": {
"obsidian-second-brain": {
"command": "node",
"args": ["/absolute/path/to/mcp-obsidian-second-brain/dist/index.js"],
"env": {
"OBSIDIAN_VAULT_PATH": "/path/to/your/obsidian/vault"
}
}
}
}
Option 3: Claude Code (CLI)
Add to ~/.claude/claude_desktop_config.json or run:
claude mcp add obsidian-second-brain \
-e OBSIDIAN_VAULT_PATH=/path/to/your/obsidian/vault \
-- npx -y @neverprepared/mcp-obsidian-second-brain
Configuration
| Environment Variable | Default | Description |
|---|---|---|
OBSIDIAN_VAULT_PATH |
~/workspaces/profiles/personal/obsidian/vaults/memory |
Path to your Obsidian vault directory |
LOG_LEVEL |
info |
Log verbosity: debug, info, warn, error (logs go to stderr) |
MIN_SHARED_TAGS |
2 |
Minimum shared tags required for auto-linking memories |
The vault directory will be created automatically on first run with the PARA folder structure.
Tools
| Tool | Description |
|---|---|
memory_store |
Create a new memory in a PARA category with optional tags and source URLs |
memory_recall |
Fetch a specific memory by ID or title |
memory_search |
Full-text search with relevance scoring, tag filters (and/or), date filters, and freshness |
memory_list |
Paginated listing with filters by PARA, tags, status, and date range |
memory_update |
Update content, tags, PARA category (moves file), or title (renames file) |
memory_archive |
Set a memory's status to archived |
memory_delete |
Permanently delete a memory (cascades to clean up backlinks) |
memory_link |
Create bidirectional [[wiki-links]] between two memories, or discover existing links |
memory_project |
Create projects with deadlines, mark complete, or list active projects |
memory_stats |
Vault health summary: counts by PARA/status, stale memories, orphans, top tags |
memory_cleanup |
Bulk list/archive/delete stale, archived, or orphaned memories (safe dry_run by default) |
Vault Structure
<OBSIDIAN_VAULT_PATH>/
├── Projects/ # Time-bound goals (stale after 30 days)
├── Areas/ # Ongoing responsibilities (stale after 90 days)
├── Resources/ # Reference material (stale after 180 days)
├── Archives/ # Inactive memories (stale after 365 days)
└── _daily/ # Daily notes (auto-appended on each store)
Each memory is a Markdown file with YAML frontmatter. Files are named by slugified title and support bidirectional [[wiki-links]].
Development
npm run dev # Run with tsx (no build required)
npm run build # Compile TypeScript to dist/
npm run typecheck # Type-check without emitting
npm test # Run all tests
npm run test:watch # Watch mode
npm run test:coverage # Coverage report
Copy .env.example to .env and set OBSIDIAN_VAULT_PATH before running locally.
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。