Log inSubmit server
Apple Fm MCP logo

Apple Fm MCP

Runs Apple’s on-device Foundation Models through MCP for private generation, summarization, classification, extraction, and status checks.

InstallRepositoryWebsite
Full README Compare alternatives Embed widget Claim this listing

Install Apple Fm MCP

Pick your client, then add it in one click or paste the config.

STDIOPython
~/Library/Application Support/Claude/claude_desktop_config.json
claude_desktop_config.json
{
  "mcpServers": {
    "apple-fm-mcp": {
      "command": "uvx",
      "args": [
        "apple-fm-mcp"
      ]
    }
  }
}

Settings → Developer → Edit Config, merge this into mcpServers, then fully quit and reopen Claude Desktop.

We couldn’t confirm it starts correctly

We ran the install command but it didn't respond within our test window — this can mean a slow first-time install rather than a real problem.

uvx apple-fm-mcp

Process exited before answering initialize.

This is an experimental automated check and can have false negatives — missing environment variables, a slow cold install, etc. It doesn’t necessarily mean something’s wrong. Last checked 5d ago.

Overview

apple-fm-mcp MCP server exposes Apple’s on-device Foundation Models through MCP tools for text generation, summarization, classification, structured extraction, and availability checks. It runs as a local stdio process on supported Apple Silicon Macs, using Apple Intelligence without API keys or cloud inference. Reach for it when short, privacy-sensitive language tasks should run locally; the README positions Claude for code generation and complex reasoning instead.

Use cases

  • Generate short text locally
  • Summarize short passages into bullets
  • Classify text into categories
  • Extract simple structured data
  • Check Apple model availability

Key features

  • On-device Apple Foundation Models
  • Free-form text generation
  • Bullet-point summarization
  • Text classification
  • Structured data extraction
  • Model availability checks

Tools5

What Apple Fm MCP lets your agent do. Expand a tool for its parameters.

Self-reported~49 tokens of context

Security & health

What it can reach and how reliably it runs — from our own checks, not the maintainer’s claims.

ActiveLast checked 1d ago

Recent health check succeeded.

7/7 checks healthy over the last 45d

Supply-chain signal

No high-severity advisories surfaced by our automated scan.

Critical 0High 0Medium 0Low 0

Scanned 9d ago via OSV.dev · apple-fm-mcp (PyPI)

Adoption & maintenance

Factual signals from GitHub, npm, and our automated checks — not a rating.

GitHub stars1Stargazers on the source repository.
Last commit4mo agoMost recent push to the default branch.
Install checkInconclusiveDidn't respond in our test window — often a slow first install.
Tools exposed5Callable tools this server registers over MCP.
Directory activity1 viewsConfig copies, upvotes, and views on AllMCPs.

How Apple Fm MCP works

What apple-fm-mcp does

The apple-fm-mcp MCP server connects MCP-compatible clients to Apple’s on-device Foundation Models. It exposes a small set of focused operations: free-form generation, bullet-point summarization, text classification, structured data extraction, and model availability checks. The underlying model is described as a 3B-parameter Apple model that runs on the Neural Engine.

Inference happens locally on the Mac. The project states that it does not require API keys, usage tokens, or cloud costs, and that data stays on the device. That makes it suitable for quick tasks involving privacy-sensitive text, provided the host meets Apple’s platform requirements.

How it works

The package starts as a local MCP process and communicates with the client through the configured command. An agent can select an operation based on the task, such as using fm_summarize for short text or fm_classify for a quick category decision. fm_status reports whether the model is available before other operations are attempted.

These figures are presented as estimates rather than guarantees.

The apple-fm-mcp MCP server is intended to complement, rather than replace, a larger general-purpose model. The supplied comparison recommends Apple’s model for quick classification, short summaries, and simple extraction, while reserving complex reasoning, code generation, and long-document summarization for Claude.

Setup and configuration

Installation uses Python’s package installer:

Terminal
pip install apple-fm-mcp

The host must run macOS 26 or later, also called Tahoe or later, on Apple Silicon hardware from the M1 through M4 generations. Apple Intelligence must be enabled in System Settings, and Python 3.10 or newer is required.

For Claude Code, add an apple-fm entry to ~/.claude/mcp.json with apple-fm-mcp as the command. Claude Desktop uses the same command in ~/Library/Application Support/Claude/claude_desktop_config.json. The configuration does not require an environment variable or API credential.

Tools and capabilities

  • fm_generate: Produces free-form text.
  • fm_summarize: Converts text into bullet-point summaries.
  • fm_classify: Assigns text to categories.
  • fm_extract: Pulls structured data from text.
  • fm_status: Checks model availability.

These operations cover lightweight local language processing rather than broad agent orchestration or advanced reasoning. The README specifically describes code generation and complex reasoning as better fits for Claude.

Limitations and notes

The apple-fm-mcp MCP server cannot run on unsupported operating systems or non-Apple-Silicon Macs according to the stated requirements. Apple Intelligence must also be enabled, so model availability depends on the local system configuration. The project does not describe remote hosting, cloud execution, or support for models other than Apple’s on-device Foundation Models.

The published setup documents Claude Code and Claude Desktop configurations. Other MCP clients may require their own method for launching a local stdio server. The repository is licensed under MIT.

Read the full README View source on GitHub

Reviews

No reviews yet — be the first to share how it worked for you.

Frequently Asked Questions about Apple Fm MCP

apple-fm-mcp is an MCP server that connects MCP clients to Apple’s on-device Foundation Models on supported Apple Silicon Macs. Its main tools generate text, summarize content into bullets, classify text, extract structured data, and check model availability. Inference runs locally through Apple Intelligence without API keys or cloud costs.

Install it with `pip install apple-fm-mcp`. The host must use macOS 26 or later, Apple Silicon, Apple Intelligence, and Python 3.10 or newer. Then configure the `apple-fm-mcp` command in the MCP client’s server configuration.

Yes, apple-fm-mcp works with Claude Desktop through its local stdio configuration. Add an `apple-fm` server entry using `apple-fm-mcp` as the command in `~/Library/Application Support/Claude/claude_desktop_config.json`.

Yes, apple-fm-mcp works with Claude Code. Add the server to `~/.claude/mcp.json` with `apple-fm-mcp` as its command, then use the exposed Foundation Models tools from the client.

apple-fm-mcp requires macOS 26 or later, an M1, M2, M3, or M4 Mac, Apple Intelligence enabled in System Settings, and Python 3.10 or newer.

Related MCP servers

All in Conversational AI All alternatives
Add a badge to your READMEShow it’s listed — live SVG, Markdown or HTML.

AllMCPs Directory Badge

Full Badge Customizer

Showcase your server listing on GitHub or your project documentation. Embed this dynamic SVG badge to highlight official listing status and live engagement.

Badge Style:
Live Dynamic SVG PreviewApple Fm MCP AllMCPs Directory Badge
Markdown (GitHub README)
[![AllMCPs](https://allmcps.com/api/badge/apple-fm-mcp?style=directory)](https://allmcps.com/mcp/apple-fm-mcp)
HTML Embed
<a href="https://allmcps.com/mcp/apple-fm-mcp"><img src="https://allmcps.com/api/badge/apple-fm-mcp?style=directory" alt="Apple Fm MCP on AllMCPs" /></a>