# Local MCP [Health: Active]

**Category:** 💬 Communication  
**Repository:** https://github.com/colibird-ai/local-mcp-releases  
**GitHub Stars:** 70  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/local-mcp-2

## Description
Mac & Windows: let ChatGPT, Claude & Cursor use your email, calendar, iMessage, Teams, files.

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

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

## Documentation

## What the Local MCP MCP server does

Local MCP MCP server connects compatible assistants with information and actions across a computer’s applications, synced data, and authorized online services. It is intended for workflows where one source is not enough, such as finding related details in email, messages, and files or combining calendar events with tasks and commitments.

Supported areas include email, calendars, contacts, messages, tasks, files, documents, browser automation, and other integrations. The README lists examples such as Apple Mail, Outlook, Microsoft Teams, Slack, WhatsApp, Signal, Google Drive, OneDrive, Microsoft 365, Notes, Reminders, Microsoft To Do, OmniFocus, Notion, and ServiceNow. Availability depends on the operating system, account access, and the current tool reference.

## How it works

The desktop application runs the tools on the user’s computer. Depending on the integration, a tool may read native application data, local caches, synced files, or a connected service that requires sign-in and network access. The assistant receives the available MCP tools and uses them to retrieve context or propose and perform supported actions.

Local MCP MCP server supports local desktop and command-line clients through a local MCP process. Web assistants use `https://www.local-mcp.com/mcp` as a custom connector with OAuth. That web connection requires the local app and Cloud Data Forwarding, which is disabled by default; without forwarding, the connector exposes only `setup_install`.

## Setup and configuration

Download and install the application for macOS 13 or later or Windows 10 or later, then select an assistant and enable the integrations you need. Grant application permissions or sign in to connected services when prompted. Restart the AI client after installation because MCP tools load when the client starts.

The npm wrapper can configure detected clients with:

```bash
npx -y local-mcp@latest setup
```

It requires Node.js. Clients that accept a standard local MCP configuration can use command `npx` with arguments `-y local-mcp@latest`. The repository also provides a Cursor plugin, a Claude Desktop extension, platform installers, and separate setup paths for web assistants.

## Tools and capabilities

Available tools vary by platform and authorization. Depending on the selected integrations, the server can provide:

- Search and context gathering across email, messages, files, documents, and calendars.
- Access to contacts, tasks, notes, reminders, and supported productivity applications.
- Supported updates or other actions in connected applications after the required permissions and authorization are granted.
- Browser automation and integrations with listed services such as Microsoft 365, Google Drive, OneDrive, Slack, and Teams.
- Cross-source workflows that let an assistant assemble information before drafting or carrying out an authorized action.

## Limitations and notes

Not every integration is available on both operating systems. Apple-native tools require macOS, and some integrations use local or synchronized data rather than a complete cloud account. Access also depends on each application’s permissions, authentication, and the assistant’s capabilities.

Linux, iOS, and Android are not supported yet. The repository contains release information, documentation, scripts, and an npm wrapper; the core binary is proprietary. Examples in the README describe possible workflows, not guaranteed autonomous results. An endpoint alone does not confirm a working connection: authorization must be completed and the client must show the available tools.

_Full upstream README: https://allmcps.com/mcp/local-mcp-2/readme_