# lanchuske/local-mcp [Health: Active]

**Category:** 💬 Communication  
**Repository:** https://github.com/lanchuske/local-mcp-releases  
**GitHub Stars:** 49  
**Views:** 3  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/lanchuske-local-mcp

## Description
Connect Claude, Cursor, Windsurf and other AI agents to macOS native apps: Mail, Calendar, Contacts, Reminders, Notes, iMessage, Finder, Safari, OmniFocus, Microsoft Teams, Outlook, OneDrive, and Office documents. 82 tools. Runs entirely on your Mac — no cloud, no tokens, no API keys.

## Claude Desktop Quick Installation
Heuristic fallback — verify the package name and runner against the repository README before running it. Uses `npx` (confidence: low):

```json
"mcpServers": {
  "local-mcp": {
    "command": "npx",
    "args": ["-y","lanchuske-local-mcp"]
  }
}
```

## Documentation

## What lanchuske/local-mcp MCP server does

lanchuske/local-mcp MCP server exposes tools for working with applications and data on a Mac through an AI client. Supported areas include Apple Mail, Calendar, Contacts, Reminders, Notes, Messages, Finder, Safari, OmniFocus, Microsoft Teams, Outlook, OneDrive, Office documents, Slack, WhatsApp, Google Drive, Zoom, ServiceNow, Signal, Notion, and NordVPN. It also includes PDF handling, stock information, and browser-style automation for websites.

The available actions cover both reading and writing. An agent can search messages, summarize email, inspect calendars, find local files, read documents, create reminders, update notes, send messages, or create events. Write operations display a preview and require confirmation before they are applied.

## How it works

The client communicates with the local process using MCP over stdio. The process then accesses macOS frameworks, application scripting interfaces, local caches, and files rather than relying exclusively on vendor APIs. Calendar access uses EventKit, Mail uses AppleScript or JXA, Contacts uses CNContactStore, and files stored by services such as OneDrive are accessed through the file system. Teams message searches can use the local IndexedDB LevelDB cache.

For desktop clients, tool execution stays on the Mac and can work without internet access for supported local data. Web AI clients cannot normally connect to localhost, so the optional Cloud Relay provides an encrypted WebSocket path to the local process. Relay use is opt-in, and the README states that relayed data is encrypted in transit and not stored.

## Setup and configuration

lanchuske/local-mcp MCP server supports macOS 13 or later on Apple Silicon and Intel Macs. The documented terminal installation command is:

```bash
curl -fsSL 'https://local-mcp.com/install?ref=github-releases' | bash
```

The installer can configure Claude Desktop, Claude Code, Cursor, Windsurf, VS Code, and Zed. After installation, the AI client must be restarted. A separate Mac app can be downloaded from the project website and opened from Applications; its menu-bar interface guides client connections without requiring manual configuration-file edits.

ChatGPT web access requires enabling Cloud Data Forwarding in the menu-bar app, creating an MCP connector pointed at `https://www.local-mcp.com/mcp`, and authorizing it with the per-machine token shown in the app. The README describes this path for supported ChatGPT plans and web access.

## Tools and capabilities

- Query and modify Mail, Calendar, Contacts, Reminders, Notes, Messages, Finder, and Safari.
- Search or work with Teams, Slack, WhatsApp, Outlook, OneDrive, Google Drive, Zoom, and Microsoft 365 data.
- Read and create Word, Excel, PowerPoint, and PDF content.
- Search local Signal data in read-only mode and use the Wacli-based WhatsApp connection with QR-code sign-in.
- Automate websites with persistent sessions and retrieve information from Notion, ServiceNow, and other listed services.

## Limitations and notes

lanchuske/local-mcp MCP server is macOS-only; Windows and Linux support is described as being on a waitlist. Some capabilities depend on the relevant application being installed, its local data being available, or additional sign-in such as WhatsApp QR-code authentication. Cloud Relay is needed for web AI clients, while local desktop clients can use the stdio connection directly. Although local operation does not require API keys or OAuth credentials, cloud-client authorization uses a token configured through the menu-bar app.

_Full upstream README: https://allmcps.com/mcp/lanchuske-local-mcp/readme_