SkillAgentSearch skills...

obsidian-connect-mcp

Query your Obsidian vault using Dataview using an MCP

Install / Use

claude mcp add joch -- npx -y github:joch/obsidian-connect-mcp

If the server publishes to npm under a different name, use that package instead — check the repo README.

About this skill
🔌

MCP Server

Model Context Protocol server

Quality Score

70/100

Supported Platforms

Claude Code
Claude Desktop

Tags

Our assessment of obsidian-connect-mcp

obsidian-connect-mcp scores 70/100 on our quality scale, 1122nd of 1,937 Development & Engineering skills we index.

Its MCP Server is 9.3 KB long, well organised into 44 sections with 8 code examples: a thorough specification that gives an agent plenty to work with.

It has 3 GitHub stars, so there is little community track record yet; judge it on its content.

Substance
29/30
Structure
20/20
Description
8/15
Adoption
3/20
Freshness
11/15

Maintenance, license and trust

  • The repository was last updated about 8 months ago. That is recent enough to be usable, but agent tooling moves fast, so check the instructions against your agent's current version.
  • Our last check on 2026-09-10 found the source still online.
  • It is released under the 0BSD license, a permissive license that allows use, modification and commercial use with attribution.
  • Its trust signals score 81/100, with 3 cautions from licensing, adoption, age or documentation. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.

Safety scan

No issues found

Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands.

Automated pattern scan on 2026-09-26. It catches known dangerous patterns, not every risk — read a skill before letting an agent act on it.

obsidian-connect-mcp compared with similar skills

All 4 of these similar skills score higher than obsidian-connect-mcp; compare them before choosing.

SkillScoreStarsUpdatedFormat
obsidian-connect-mcp (this skill)by joch7038mo agoMCP Server
Agent-Reachby Panniantong10085.4k10d agoCLAUDE.md
headroomby headroomlabs-ai10073.8ktodayCLAUDE.md
rufloby ruvnet10073.3k1d agoCLAUDE.md
CowAgentby zhayujie10047.1ktodayCLAUDE.md

Frequently asked questions

How do I install obsidian-connect-mcp?
Run claude mcp add joch -- npx -y github:joch/obsidian-connect-mcp. The install tabs above show the steps for each supported agent.
Which AI agents does obsidian-connect-mcp work with?
It is written for Claude Code and Claude Desktop, as a MCP Server file. Other agents that read the same format can often use it too.
Is obsidian-connect-mcp safe to use?
Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. It is 0BSD-licensed and scores 81/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
Is obsidian-connect-mcp still maintained?
The repository was last updated about 8 months ago. That is recent enough to be usable, but agent tooling moves fast, so check the instructions against your agent's current version.

Connect MCP

An Obsidian plugin that runs an MCP (Model Context Protocol) server, enabling AI agents like Claude to access your vault - read, edit, search notes, and run Dataview queries.

Architecture

AI Agent (Claude Code/Desktop)
    ↓ (via mcp-remote bridge)
Obsidian Plugin (Express + MCP Server on port 27124)
    ↓
Obsidian Vault

The MCP server runs directly inside the Obsidian plugin - no separate server process needed.

Prerequisites

Installation

Option A: Install via BRAT (Recommended)

  1. Install the BRAT plugin from Community Plugins
  2. Open BRAT settings and click "Add Beta Plugin"
  3. Enter: joch/obsidian-connect-mcp
  4. Enable "Connect MCP" in Community Plugins

Option B: Manual Installation

  1. Build the plugin:

    npm install
    npm run build
    
  2. Copy the plugin files to your Obsidian vault:

    mkdir -p /path/to/vault/.obsidian/plugins/connect-mcp
    cp main.js manifest.json styles.css /path/to/vault/.obsidian/plugins/connect-mcp/
    
  3. In Obsidian, go to Settings → Community plugins → Enable "Connect MCP"

  4. Configure the plugin:

    • Go to Settings → Connect MCP
    • Click "Generate" to create an API Key
    • Optionally change the port (default: 27124)
    • Click "Start Server" or enable auto-start

2. Configure Claude Code/Desktop

Add this to your Claude Code MCP settings:

{
  "mcpServers": {
    "obsidian": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "http://localhost:27124/mcp",
        "--header",
        "Authorization:${AUTH}"
      ],
      "env": {
        "AUTH": "Bearer YOUR_API_KEY"
      }
    }
  }
}

Replace YOUR_API_KEY with the key generated in Obsidian settings.

Available MCP Tools

Vault Tools

| Tool | Description | |------|-------------| | vault_list | List files and folders in the vault | | vault_read | Read a note's content and frontmatter | | vault_create | Create a new note | | vault_update | Replace entire file content | | vault_delete | Delete a note (moves to trash) | | vault_search | Search notes by content |

Edit Tools

| Tool | Description | |------|-------------| | vault_edit | Find and replace text using fuzzy matching | | vault_edit_line | Insert or replace content at a specific line | | vault_patch | Edit a specific section: heading, block, or frontmatter |

Graph Tools

| Tool | Description | |------|-------------| | graph_info | Get link statistics for a note | | graph_links | Get backlinks and forward links for a note |

Dataview Tools

| Tool | Description | |------|-------------| | dataview_query | Execute a Dataview DQL query |

Active Tools

| Tool | Description | |------|-------------| | active_note | Get the currently open note in Obsidian |

Command Tools

| Tool | Description | |------|-------------| | command_list | List available Obsidian commands (filter by plugin name) | | command_execute | Execute any Obsidian command by ID (Templater, MetaBind, etc.) |

Note: Command tools must be enabled in plugin settings ("Allow command execution").

Templater Integration

The command_execute tool can run Templater templates as commands, enabling AI agents to trigger your automation scripts (e.g., "consume wine bottle", "complete project", "mark article as read").

Exposing Templates as Commands

By default, Templater templates are only accessible via the UI modal. To make them available as commands:

  1. Open Obsidian Settings → Templater
  2. Scroll to Template Hotkeys
  3. Click Add new template hotkey for each template you want to expose
  4. Select the template file (e.g., apps/templater/wine/consume.md)
  5. Restart Obsidian to register the new commands

Each template gets two commands:

  • templater-obsidian:path/to/template.md - Run on active file ("Insert")
  • templater-obsidian:create-path/to/template.md - Create new file from template

Example Usage

# List available wine commands
command_list with filter: "wine"

# Execute consume template on active bottle note
command_execute with commandId: "templater-obsidian:apps/templater/wine/consume.md"

Recommended Templates to Expose

Add templates that perform actions on existing notes:

  • Status changes (consume bottle, complete project, mark as read)
  • Property updates (cycle status, change priority)
  • File operations (archive, move to folder)

Templates that require user input (modals, prompts) may not work well via MCP.

MCP Resources

| URI | Description | |-----|-------------| | obsidian://vault-info | Vault name, file counts, plugin status | | obsidian://dataview-reference | DQL syntax quick reference |

MCP Prompts

Prompts let you give AI agents context about your vault that they can discover and use automatically. Create markdown files in the prompts folder (configurable in settings) to help agents understand your vault structure, conventions, and useful queries.

Why Use Prompts?

When an AI agent connects to your vault, it has no idea how you've organized things. Prompts solve this by providing:

  • Vault structure - Describe your folder hierarchy and what goes where
  • Dataview queries - Pre-built DQL queries for common searches in your vault
  • Naming conventions - Explain your file naming patterns and metadata fields
  • Workflows - Document how you use tags, links, or specific note types

Prompt Format

---
description: Brief description the agent sees when listing prompts
---

# My Vault Guide

## Folder Structure
- `projects/` - Active project notes with status frontmatter
- `archive/` - Completed projects
- `daily/` - Daily notes in YYYY-MM-DD format

## Useful Queries

Find incomplete tasks:
\`\`\`dataview
TASK FROM "projects" WHERE !completed
\`\`\`

Recent notes:
\`\`\`dataview
LIST FROM "" WHERE file.mtime > date(today) - dur(7 days) SORT file.mtime DESC
\`\`\`

## Conventions
- Projects use `status` field: active, waiting, done
- People notes are in `people/` with `birthday` and `company` fields

The filename (without .md) becomes the prompt name that agents can request.

Security Features

API Key Authentication

All requests require a valid API key in the Authorization header.

Read-Only Mode

Enable in settings to block all write operations (create, update, delete).

.mcpignore File

Create a .mcpignore file in your vault root to block access to sensitive paths:

# Block private folders
private/
journal/

# Block specific files
secrets.md

Uses gitignore-style patterns.

Example DQL Queries

-- List notes with a tag
LIST FROM #project

-- Table of tasks
TABLE file.name, due, status FROM "Tasks" WHERE !completed

-- Notes modified today
LIST FROM "" WHERE file.mday = date(today)

-- Count notes by folder
TABLE length(rows) as Count FROM "" GROUP BY file.folder

Remote Access (Cloud Agents)

To securely expose your MCP server to AI agents running in the cloud, use Cloudflare Tunnel. This provides HTTPS encryption without opening ports on your firewall.

Setup Cloudflare Tunnel

  1. Install cloudflared:

    # macOS
    brew install cloudflared
    
    # Windows
    winget install Cloudflare.cloudflared
    
    # Linux (Debian/Ubuntu)
    curl -L https://pkg.cloudflare.com/cloudflare-main.gpg | sudo tee /usr/share/keyrings/cloudflare-archive-keyring.gpg
    echo "deb [signed-by=/usr/share/keyrings/cloudflare-archive-keyring.gpg] https://pkg.cloudflare.com/cloudflared $(lsb_release -cs) main" | sudo tee /etc/apt/sources.list.d/cloudflared.list
    sudo apt update && sudo apt install cloudflared
    
  2. Create a quick tunnel (no Cloudflare account needed):

    cloudflared tunnel --url http://localhost:27124
    

    This outputs a URL like https://random-words.trycloudflare.com

  3. Update your Claude Code MCP settings to use the tunnel URL:

    {
      "mcpServers": {
        "obsidian": {
          "command": "npx",
          "args": [
            "mcp-remote",
            "https://random-words.trycloudflare.com/mcp",
            "--header",
            "Authorization:${AUTH}"
          ],
          "env": {
            "AUTH": "Bearer YOUR_API_KEY"
          }
        }
      }
    }
    

Persistent Tunnel (Optional)

For a stable URL, create a named tunnel with a Cloudflare account:

  1. Login to Cloudflare:

    cloudflared tunnel login
    
  2. Create a named tunnel:

    cloudflared tunnel create obsidian-mcp
    
  3. Configure the tunnel (create ~/.cloudflared/config.yml):

    tunnel: obsidian-mcp
    credentials-file: ~/.cloudflared/<tunnel-id>.json
    
    ingress:
      - hostname: obsidian-mcp.yourdomain.com
        service: http://localhost:27124
      - service: http_status:404
    
  4. Add DNS record:

    cloudflared tunnel route dns obsidian-mcp obsidian-mcp.yourdomain.com
    
  5. Run the tunnel:

    cloudflared tunnel run obsidian-mcp
    

Verifying the Setup

# Health check (no auth required)
curl http://localhost:27124/health

# MCP info
curl http://localhost:27124/mcp

# Via tunnel (if using remote access)
curl https://your-tunnel-url.trycloudflare.com/health

Development

npm run dev  # Watch mode

License

MIT

Related Skills

View on GitHub
GitHub Stars3
CategoryDevelopment
Updated7mo ago
Forks0

Languages

TypeScript

Trust signals

81/100

From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.

3 low