Sponsored by Deepsite.site

Obsidian Md Mcp Via Rest Api

Created By
Oleksandr Kucherenko8 months ago
Dockerized MCP Server connected to Obsidian Vault with Local REST API. (rewritten completely with BUN to make it faster and more lightweight)
Content

mcp-obsidian

Deployed / Usage

Ask DeepWiki NPM Downloads Docker Hub obsidian-mcp Docker Hub obsidian-vnc

CI/CD Status

NPM (npmjs.org) NPM (GitHub)

Docker (GitHub) Docker (Docker Hub)

Screenshots Cleanup


MCP Tools & Resources

This MCP server exposes the following tools and resources to AI assistants:

Tools

ToolDescriptionParameters
get_note_contentRetrieve content and metadata of an Obsidian notefilePath (string) - Path to the note
obsidian_searchSearch notes using a query stringquery (string) - Search query
obsidian_semantic_searchSemantic search for notesquery (string) - Search query

Resources

ResourceURI PatternDescription
Obsidian Noteobsidian://{path}Access notes via URI (e.g., obsidian://Daily/2025-01-16.md)

Quick Test

# 1. Set your Obsidian API key
export OBSIDIAN_API_KEY="your-obsidian-rest-api-key"

# 2. Add MCP server and test (Claude Code)
claude mcp add obsidian -- bunx -y @oleksandrkucherenko/mcp-obsidian
claude "Search my Obsidian vault for monitoring tools, summarize findings"

# 2. Alternative: Codex CLI
codex mcp add obsidian --command "bunx -y @oleksandrkucherenko/mcp-obsidian"
codex "Find notes about logging frameworks and create a comparison table"

Example use-case: "Find all tools in my Obsidian vault for tracking logs and metrics, make a summary report" — the AI searches your vault, finds notes about OpenTelemetry, Datadog, Prometheus, etc., and generates a structured summary.

For other CLI tools (Gemini, OpenCode, Kilo Code, Copilot), see Manual Testing Guide.

Gemini CLI Example

Configure MCP

Use API_URLS for automatic failover and self-healing. The server tests all URLs in parallel, selects the fastest one, and automatically reconnects on failure.

{
  "mcpServers": {
    "obsidian": {
      "command": "docker",
      "args": [
        "run",
        "--name", "mcp-obsidian",
        "--rm",
        "-i",  // Keep STDIN open for stdio transport
        "-p", "3000:3000",
        "-e", "API_KEY",
        "-e", "API_URLS",
        "-e", "DEBUG", // for logs
        "ghcr.io/oleksandrkucherenko/obsidian-mcp:latest"
      ],
      "env": {
        "API_KEY": "<secret_key>",
        // JSON array - automatically tests and selects fastest URL
        "API_URLS": "[\"https://127.0.0.1:27124\",\"https://172.26.32.1:27124\",\"https://host.docker.internal:27124\"]",
        "DEBUG": "mcp:*"
      }
    }
  }
}

Self-Healing Features:

  • ✅ Parallel URL testing on startup
  • ✅ Automatic selection of fastest URL
  • ✅ Health monitoring every 30 seconds
  • ✅ Automatic failover on connection loss
  • ✅ Exponential backoff to prevent thrashing

Available transports:

  • stdio - Standard input/output (default, best for local MCP clients)
  • http - HTTP JSON-RPC with SSE streaming (best for remote access)

WSL2 Example:

# Automatically determine WSL gateway IP
export WSL_GATEWAY_IP=$(ip route show | grep -i default | awk '{ print $3}')

# Configure with multiple fallback URLs
API_URLS='["https://127.0.0.1:27124", "https://'$WSL_GATEWAY_IP':27124", "https://host.docker.internal:27124"]'

HTTP Transport Configuration

The MCP server supports HTTP transport for remote access with automatic URL failover:

{
  "mcpServers": {
    "obsidian-http": {
      "command": "docker",
      "args": [
        "run",
        "--name", "mcp-obsidian-http",
        "--rm",
        "-p", "3000:3000",
        "-e", "API_KEY",
        "-e", "API_URLS",
        "-e", "MCP_HTTP_PATH",
        "ghcr.io/oleksandrkucherenko/obsidian-mcp:latest"
      ],
      "env": {
        "API_KEY": "<secret_key>",
        "API_URLS": "[\"https://127.0.0.1:27124\",\"https://172.26.32.1:27124\",\"https://host.docker.internal:27124\"]",
        "MCP_HTTP_PATH": "/mcp" // endpoint path (default is: /mcp)
      }
    }
  }
}

Decoupled Configuration with Authentication

# Automatically determine WSL gateway IP
export WSL_GATEWAY_IP=$(ip route show | grep -i default | awk '{ print $3}')

# Configure with multiple fallback URLs
API_URLS='["https://127.0.0.1:27124", "https://'$WSL_GATEWAY_IP':27124", "https://host.docker.internal:27124"]'

# run MCP server on docker separately from IDE
docker run --name mcp-obsidian-http --rm \
  -p 3000:3000 \
  -e API_KEY="<secret_key>" \
  -e API_URLS="${API_URLS}" \
  -e MCP_HTTP_TOKEN=<your-secret-token-here> \
  ghcr.io/oleksandrkucherenko/obsidian-mcp:latest
{
  "mcpServers": {
    "obsidian": {
      "type": "streamable-http",
      "url": "http://localhost:3000/mcp",
      "headers": {
        "Authorization": "Bearer <your-secret-token-here>"
      }      
    }
  }
}

Clients must include the Authorization header:

Authorization: Bearer your-secret-token-here

Stdio Transport Configuration

For local development with stdio transport (default):

{
  "mcpServers": {
    "obsidian": {
      "command": "docker",
      "args": [
        "run",
        "--name", "mcp-obsidian-windsurf",
        "--interactive",
        "--rm",
        "-e", "API_KEY",
        "-e", "API_URLS",
        "-e", "DEBUG",
        "ghcr.io/oleksandrkucherenko/obsidian-mcp:latest"
      ],
      "env": {
        "API_KEY": "<secret_key>",
        "API_URLS": "[\"https://127.0.0.1:27124\",\"https://172.26.32.1:27124\"]",
        "DEBUG": "mcp:*" // default: disabled logs
      }
    }
  }
}
  • --rm - Automatically remove the container and its associated anonymous volumes when it exits

  • -i, --interactive - Keep STDIN open

  • -e, --env - Set environment variables

  • --name string - Assign a name to the container

  • -p, --publish - Publish container port to host

  • NPM Package Releases

  • Docker Image Releases

Legacy Single-URL Configuration

For backward compatibility, you can still use single-URL configuration with API_HOST and API_PORT:

{
  "mcpServers": {
    "obsidian": {
      "command": "docker",
      "args": [
        "run",
        "--name", "mcp-obsidian",
        "--rm",
        "-i",
        "-e", "API_KEY",
        "-e", "API_HOST",
        "-e", "API_PORT",
        "ghcr.io/oleksandrkucherenko/obsidian-mcp:latest"
      ],
      "env": {
        "API_KEY": "<secret_key>",
        "API_HOST": "https://172.26.32.1",  // single URL without failover
        "API_PORT": "27124"
      }
    }
  }
}

Note: Single-URL configuration does not provide automatic failover or health monitoring. Use API_URLS for production deployments.

Health Endpoint

When HTTP transport is enabled, the server exposes a health check endpoint at /health:

curl http://localhost:3000/health

Response:

{
  "status": "healthy",
  "timestamp": "2025-01-12T12:00:00.000Z",
  "transport": "http",
  "authEnabled": false
}

For comprehensive health status including Obsidian API connection and all transports, you can use the getHealthStatus() function which returns:

{
  "healthy": true,
  "obsidian": {
    "connected": true,
    "url": "https://obsidian:27124",
    "lastCheck": 1705065600000
  },
  "transports": {
    "stdio": { "running": true, "enabled": true },
    "http": { "running": true, "enabled": true }
  },
  "uptime": 3600,
  "timestamp": 1705065600000
}

CLI Tools Configuration

This section shows how to configure popular AI CLI tools to use the MCP Obsidian server.

# MacOs or Linux
curl -fsSL https://bun.sh/install | bash

# Windows
powershell -c "irm bun.sh/install.ps1 | iex"

Claude Code CLI

Docker:

# Create mcp.json configuration
cat > mcp.json << 'EOF'
{
  "mcpServers": {
    "obsidian": {
      "command": "docker",
      "args": ["run", "--rm", "-i", "-e", "API_KEY", "-e", "API_URLS", "ghcr.io/oleksandrkucherenko/obsidian-mcp:latest"],
      "env": {
        "API_KEY": "<your-obsidian-api-key>",
        "API_URLS": "[\"https://host.docker.internal:27124\"]"
      }
    }
  }
}
EOF

# Run Claude with MCP config
claude --mcp-config ./mcp.json

NPX/Bunx:

cat > mcp.json << 'EOF'
{
  "mcpServers": {
    "obsidian": {
      "command": "bunx",
      "args": ["-y", "@oleksandrkucherenko/mcp-obsidian"],
      "env": {
        "API_KEY": "<your-obsidian-api-key>",
        "API_URLS": "[\"https://127.0.0.1:27124\"]"
      }
    }
  }
}
EOF

claude --mcp-config ./mcp.json

Gemini CLI

Docker:

gemini mcp add \
  -e API_KEY=<your-obsidian-api-key> \
  -e API_URLS='["https://host.docker.internal:27124"]' \
  obsidian \
  docker run --rm -i ghcr.io/oleksandrkucherenko/obsidian-mcp:latest

NPX/Bunx:

gemini mcp add \
  -e API_KEY=<your-obsidian-api-key> \
  -e API_URLS='["https://127.0.0.1:27124"]' \
  obsidian \
  bunx -y @oleksandrkucherenko/mcp-obsidian

HTTP Transport (remote server):

gemini mcp add --transport http obsidian-http http://localhost:3000/mcp

List and manage servers:

gemini mcp list
gemini mcp remove obsidian

OpenCode CLI

Create opencode.json in your project root:

Docker:

{
  "mcp": {
    "obsidian": {
      "type": "local",
      "command": ["docker", "run", "--rm", "-i",
        "-e", "API_KEY", "-e", "API_URLS",
        "ghcr.io/oleksandrkucherenko/obsidian-mcp:latest"],
      "environment": {
        "API_KEY": "{env:API_KEY}",
        "API_URLS": "[\"https://host.docker.internal:27124\"]"
      },
      "enabled": true
    }
  }
}

NPX/Bunx:

{
  "mcp": {
    "obsidian": {
      "type": "local",
      "command": ["bunx", "-y", "@oleksandrkucherenko/mcp-obsidian"],
      "environment": {
        "API_KEY": "{env:API_KEY}",
        "API_URLS": "[\"https://127.0.0.1:27124\"]"
      },
      "enabled": true
    }
  }
}

HTTP Transport:

{
  "mcp": {
    "obsidian-http": {
      "type": "remote",
      "url": "http://localhost:3000/mcp",
      "enabled": true
    }
  }
}

Kilo Code CLI

Create .kilocode/mcp.json in your project or ~/.kilocode/cli/global/settings/mcp_settings.json globally:

Docker:

{
  "mcpServers": {
    "obsidian": {
      "command": "docker",
      "args": ["run", "--rm", "-i", "-e", "API_KEY", "-e", "API_URLS",
        "ghcr.io/oleksandrkucherenko/obsidian-mcp:latest"],
      "env": {
        "API_KEY": "<your-obsidian-api-key>",
        "API_URLS": "[\"https://host.docker.internal:27124\"]"
      }
    }
  }
}

NPX/Bunx:

{
  "mcpServers": {
    "obsidian": {
      "command": "bunx",
      "args": ["-y", "@oleksandrkucherenko/mcp-obsidian"],
      "env": {
        "API_KEY": "<your-obsidian-api-key>",
        "API_URLS": "[\"https://127.0.0.1:27124\"]"
      }
    }
  }
}

HTTP Transport:

{
  "mcpServers": {
    "obsidian-http": {
      "type": "streamable-http",
      "url": "http://localhost:3000/mcp",
      "headers": {
        "Authorization": "Bearer <your-token>"
      }
    }
  }
}

Codex CLI

Docker:

# Register MCP server
codex mcp add obsidian \
  --command "docker run --rm -i -e API_KEY -e API_URLS ghcr.io/oleksandrkucherenko/obsidian-mcp:latest" \
  --env API_KEY=<your-obsidian-api-key> \
  --env 'API_URLS=["https://host.docker.internal:27124"]'

NPX/Bunx:

codex mcp add obsidian \
  --command "bunx -y @oleksandrkucherenko/mcp-obsidian" \
  --env API_KEY=<your-obsidian-api-key> \
  --env 'API_URLS=["https://127.0.0.1:27124"]'

GitHub Copilot CLI

Create ~/.copilot/mcp-config.json (or .copilot/mcp-config.json in repo root):

Docker:

{
  "mcpServers": {
    "obsidian": {
      "type": "local",
      "command": "docker",
      "args": ["run", "--rm", "-i", "-e", "API_KEY", "-e", "API_URLS",
        "ghcr.io/oleksandrkucherenko/obsidian-mcp:latest"],
      "env": {
        "API_KEY": "${OBSIDIAN_API_KEY}",
        "API_URLS": "[\"https://host.docker.internal:27124\"]"
      },
      "tools": ["*"]
    }
  }
}

NPX/Bunx:

{
  "mcpServers": {
    "obsidian": {
      "type": "local",
      "command": "bunx",
      "args": ["-y", "@oleksandrkucherenko/mcp-obsidian"],
      "env": {
        "API_KEY": "${OBSIDIAN_API_KEY}",
        "API_URLS": "[\"https://127.0.0.1:27124\"]"
      },
      "tools": ["*"]
    }
  }
}

Note: Copilot CLI v0.0.340+ requires ${VAR} syntax for environment variable expansion. Set OBSIDIAN_API_KEY in your shell before running.

Quick Reference

CLI ToolConfig FileDocker SupportHTTP Transport
Claude Codemcp.json
Geminisettings.json
OpenCodeopencode.json
Kilo Code.kilocode/mcp.json
CodexCLI commands
Copilot~/.copilot/mcp-config.json

For detailed testing and verification, see Manual Testing Guide.

Setup and Troubleshooting

Setup

Obsidian Local REST API Setup

This setting will allow you to connect to the Local REST API from any network interface (not only localhost, which is critical for WSL2 setup).

  • Copy the API Key from Obsidian Settings; you will need it for the MCP configuration.

  • Verify that the Obsidian Local REST API is running and accessible from your machine.

  • Next Step is always to verify the network setup on your machine (firewall rules, etc).

Verify that the Obsidian REST API is running (Windows Host, MacOS, Linux)

Run in Windows CMD terminal:

# windows CMD, verify that port is listening (that rest api is running)
netstat -an | findstr 27124
# Expected output:
#   TCP    0.0.0.0:27124           0.0.0.0:0               LISTENING

# Verify that Obsidian Local REST API is working
curl --insecure https://localhost:27124
wget --no-check-certificate -S https://localhost:27124
http --verify=no https://localhost:27124

Expected REST API response:

{
  "status": "OK",
  "manifest": {
    "id": "obsidian-local-rest-api",
    "name": "Local REST API",
    "version": "3.2.0",
    "minAppVersion": "0.12.0",
    "description": "Get, change or otherwise interact with your notes in Obsidian via a REST API.",
    "author": "Adam Coddington",
    "authorUrl": "https://coddingtonbear.net/",
    "isDesktopOnly": true,
    "dir": ".obsidian/plugins/obsidian-local-rest-api"
  },
  "versions": {
    "obsidian": "1.8.10",
    "self": "3.2.0"
  },
  "service": "Obsidian Local REST API",
  "authenticated": false
}

WSL2, Docker hosted on Ubuntu

graph LR
    subgraph "Windows Machine"
      obs("Obsidian Application")
    
      subgraph "WSL2"
        subgraph "Ubuntu"
          subgraph "Docker"
            mcp("mcp-obsidian:latest")
          end
        end
      end

      firewall(["Windows Firewall"]) -->|27124| obs

      mcp -->|https://$WSL_GATEWAY_IP:27124| firewall

      IDE -.->|MCP Server Tools| mcp
    end

Run inside the WSL2 Ubuntu Terminal:

export WSL_GATEWAY_IP=$(ip route show | grep -i default | awk '{ print $3}')
echo $WSL_GATEWAY_IP # expected something like: 172.26.32.1

# Verify that Obsidian Local REST API is working 
curl --insecure https://$WSL_GATEWAY_IP:27124
wget --no-check-certificate -S https://$WSL_GATEWAY_IP:27124
http --verify=no https://$WSL_GATEWAY_IP:27124

Verify Windows Firewall

Run GUI and Setup Manual The Rules:

# Windows Defender Firewall / Inbound Rules. Press Win+R and type WF.msc or firewall.cpl
WF.msc
firewall.cpl # and then press 'Advanced settings'

Or Run in Windows PowerShell as Administrator:

# Add firewall rule to allow port 27124 (Run in Admin PowerShell)
New-NetFirewallRule -DisplayName "WSL2 Obsidian REST API" -Direction Inbound -LocalPort 27123,27124 -Protocol TCP -Action Allow

Or Run in Windows CMD terminal:

# check firewall rules (CMD) that manage 27124 port
netsh advfirewall firewall show rule name=all | findstr /C:"Rule Name" /C:"LocalPort" /C:"RemotePort" | findstr /C:"27124"

# display rules that has WSL2 keyword in own name
netsh advfirewall firewall show rule name=all | grep -A 13 WSL2

# display rule definition by port number (4 line after, 9 lines before)
netsh advfirewall firewall show rule name=all | grep -A 4 -B 9 27124

Disable/Enable Firewall

Execute in Windows PowerShell as Administrator:

# Temporarily turn off firewall (for testing ONLY, not recommended for regular use)
Set-NetFirewallProfile -Profile Domain,Public,Private -Enabled False

# Restore Firewall state
Set-NetFirewallProfile -Profile Domain,Public,Private -Enabled True

Verify Connectivity on BusyBox Container

These steps allow us to confirm that the network setup is correct and the container can connect to the Local REST API.

Execute inside the WSL2 Ubuntu terminal:

export WSL_GATEWAY_IP=$(ip route | grep default | awk '{print $3}')
echo "Windows host IP from WSL2: $WSL_GATEWAY_IP"
# Output:
#   Windows host IP from WSL2: 172.26.32.1

# run docker container to verify the connectivity from Docker inside
docker run --rm -it --network=host busybox sh

# inside the container run:
which wget
# /bin/wget

export WINDOWS_HOST_IP="172.26.32.1"
echo $WINDOWS_HOST_IP
# 172.26.32.1

# try to connect to the Local REST API
wget -qO- --no-check-certificate "https://$WINDOWS_HOST_IP:27124"
wget -qO- --no-check-certificate https://172.26.32.1:27124

Dockerized Obsidian

Obsidian Dockerized

Server Config

{
  "mcpServers": {
    "obsidian": {
      "command": "docker",
      "args": [
        "run",
        "--rm",
        "-i",
        "-e",
        "API_KEY",
        "-e",
        "API_URLS",
        "ghcr.io/oleksandrkucherenko/obsidian-mcp:latest"
      ],
      "env": {
        "API_KEY": "<your-obsidian-api-key>",
        "API_URLS": "[\"https://host.docker.internal:27124\"]"
      }
    }
  }
}
Recommend Servers
TraeBuild with Free GPT-4.1 & Claude 3.7. Fully MCP-Ready.
Visual Studio Code - Open Source ("Code - OSS")Visual Studio Code
Amap Maps高德地图官方 MCP Server
WindsurfThe new purpose-built IDE to harness magic
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。
AiimagemultistyleA Model Context Protocol (MCP) server for image generation and manipulation using fal.ai's Stable Diffusion model.
TimeA Model Context Protocol server that provides time and timezone conversion capabilities. This server enables LLMs to get current time information and perform timezone conversions using IANA timezone names, with automatic system timezone detection.
BlenderBlenderMCP connects Blender to Claude AI through the Model Context Protocol (MCP), allowing Claude to directly interact with and control Blender. This integration enables prompt assisted 3D modeling, scene creation, and manipulation.
MCP AdvisorMCP Advisor & Installation - Use the right MCP server for your needs
DeepChatYour AI Partner on Desktop
Playwright McpPlaywright MCP server
MiniMax MCPOfficial MiniMax Model Context Protocol (MCP) server that enables interaction with powerful Text to Speech, image generation and video generation APIs.
EdgeOne Pages MCPAn MCP service designed for deploying HTML content to EdgeOne Pages and obtaining an accessible public URL.
Zhipu Web SearchZhipu Web Search MCP Server is a search engine specifically designed for large models. It integrates four search engines, allowing users to flexibly compare and switch between them. Building upon the web crawling and ranking capabilities of traditional search engines, it enhances intent recognition capabilities, returning results more suitable for large model processing (such as webpage titles, URLs, summaries, site names, site icons, etc.). This helps AI applications achieve "dynamic knowledge acquisition" and "precise scenario adaptation" capabilities.
Serper MCP ServerA Serper MCP Server
Tavily Mcp
Howtocook Mcp基于Anduin2017 / HowToCook (程序员在家做饭指南)的mcp server,帮你推荐菜谱、规划膳食,解决“今天吃什么“的世纪难题; Based on Anduin2017/HowToCook (Programmer's Guide to Cooking at Home), MCP Server helps you recommend recipes, plan meals, and solve the century old problem of "what to eat today"
ChatWiseThe second fastest AI chatbot™
CursorThe AI Code Editor
Y GuiA web-based graphical interface for AI chat interactions with support for multiple AI models and MCP (Model Context Protocol) servers.
Jina AI MCP ToolsA Model Context Protocol (MCP) server that integrates with Jina AI Search Foundation APIs.