Daemon OS

Your AI agent can write code, run tests, search files. But it can't click a button, send an email, or fill out a form. It lives inside a chat box.

Daemon OS changes that. One install, and any AI agent can see and operate every app on your Mac.

You:     "Send an email to [email protected] about the Q4 report"
Agent:   daemon_run recipe:"gmail-send" params:{recipient, subject, body}
         → Compose opens, fields fill, email sends. Done.

Install

brew install hammadxcm/daemon-os/daemon-os
daemon setup

That's it. daemon setup handles permissions, MCP configuration, and recipe installation.

Why Daemon OS?

Other computer-use tools take screenshots and guess what's on screen. Daemon OS reads the macOS accessibility tree — structured, labeled data about every element in every app. No vision model needed. No pixel-guessing.

And when it figures out a workflow, it saves it. Other tools repeat the same expensive reasoning every time.

• Self-learning — A frontier model figures out the workflow once. A small model runs it forever.
• Transparent — Recipes are JSON. Read every step before running. No black box.
• Native — Accessibility tree, not screenshots. Structured data, not pixel coordinates.
• Any app — Not just browsers. Slack, Finder, Messages — anything on your Mac.
• Local — Your data never leaves your machine.
• Open — MCP protocol. Works with Claude Code, Cursor, VS Code, or any MCP client.

How It Works

Daemon OS connects to your AI agent through MCP and gives it 20 tools to see and operate your Mac. It reads the macOS accessibility tree for structured data about every app, and takes screenshots when visual context is needed. Click, type, scroll, press keys, manage windows. Any app, not just browsers.

You:     "Download the latest paper on chain-of-thought prompting from arXiv"
Agent:   daemon_run recipe:"arxiv-download" params:{query:"chain of thought prompting"}
         → Navigates to arXiv, searches, opens PDF, downloads to Desktop. Done.

Works with Claude Code, Cursor, VS Code, or anything that speaks MCP.

Tech Stack

20 Tools

Recipes

When your agent figures out a workflow, it saves it as a recipe. A recipe is a JSON file with steps, parameters, and wait conditions. Transparent and auditable.

A frontier model figures out the workflow once. A small model runs it forever.

# One command sends an email
daemon_run recipe:"gmail-send" params:{"recipient":"[email protected]","subject":"Hello","body":"World"}

# 7 steps, 30 seconds, 100% reliable

Bundled recipes: gmail-send, slack-send, finder-create-folder, arxiv-download

• Recipes are just JSON. Read every step before running.
• Share with your team. One person learns the workflow, everyone benefits.
• Chain recipes together. The agent knows when to call what.
• Write once with Claude or GPT-4. Run forever with Haiku.

MCP Client Configuration

daemon setup auto-configures your MCP client. For manual configuration, add to your client config:

{
  "mcpServers": {
    "daemon-os": {
      "type": "stdio",
      "command": "/usr/local/bin/daemon",
      "args": ["mcp"]
    }
  }
}

{
  "mcpServers": {
    "daemon-os": {
      "type": "stdio",
      "command": "/usr/local/bin/daemon",
      "args": ["mcp"]
    }
  }
}

{
  "servers": {
    "daemon-os": {
      "type": "stdio",
      "command": "/usr/local/bin/daemon",
      "args": ["mcp"]
    }
  }
}

{
  "mcpServers": {
    "daemon-os": {
      "type": "stdio",
      "command": "/usr/local/bin/daemon",
      "args": ["mcp"]
    }
  }
}

CLI Commands

Diagnostics

$ daemon doctor

  ✓ Accessibility: granted
  ✓ Screen Recording: granted
  ✓ Processes: 1 daemon MCP process
  ✓ MCP Config: daemon-os connected
  ✓ Recipes: 4 installed
  ✓ AX Tree: 12/12 apps readable

  All checks passed. Daemon OS is healthy.

Architecture

graph TD
    A["AI Agent (Claude Code / Cursor / VS Code)"] -->|MCP Protocol - stdio| B[Daemon OS MCP Server]
    B --> C["Perception (7 tools)"]
    B --> D["Actions (7 tools)"]
    B --> E[Wait Manager]
    B --> F[Recipe Engine]
    C --> G[AXorcist]
    D --> G
    E --> G
    F --> H[Recipe Store]

Build From Source

git clone https://github.com/hammadxcm/daemon-os.git
cd daemon-os
swift build
.build/debug/daemon setup

Requires Swift 6.2+ and macOS 14+.

Project Stats

~4,500 lines of Swift · 161 tests · 23 test suites

Credits

Built on AXorcist by @steipete — the macOS accessibility engine that makes everything possible.

Documentation

• Architecture — module design, data flow, and key decisions
• Setup — prerequisites, installation, and permissions
• Tools — complete reference for all 20 tools
• Recipes — writing and managing replayable workflows
• Contributing — development setup, testing, and PR process
• Troubleshooting — common errors and fixes

Contributing

See docs/Contributing.md. We need recipes for more apps, testing on different setups, and bug reports. If you're building AI agents that do real things, this is the project.

License

MIT