Our take
Inactiveno commits in over a year- Activity
- 19
- Community
- 15
- Issues
- 21
- Pull requests
- 20
+2 stars in the last 30 days
- Fewer contributors than 86% of the projects we track
This is a focused way to expose remote Mac desktop controls to MCP clients, but its inactive status and narrow authentication support make it a cautious choice for anything security-sensitive or business-critical.
Good fit if
- You already use Docker and an MCP client, and need an agent to interact with a Mac over Screen Sharing.
- You can test the documented protocol and security limitations on a machine you are authorized to access.
- MIT licensing works for your intended use, and you can tolerate maintaining or adapting the project yourself.
Look elsewhere if
- You need an actively maintained integration: there have been no commits in over a year and no release in the past year.
- You require broader or modern authentication support; the project documents support only for protocol 30.
- You need a mature project with a broader maintainer base: most contributions come from one of its two contributors, and recent issue and pull-request activity is absent.
All health signals
| Last commit | 2025-06-10 (1.3 years ago) |
|---|---|
| Commits, last 90 days | 0 |
| Releases, last 12 months | 0 (latest v0.0.2, 2025-04-16) |
| Contributors | 2 (top contributor: 96% of commits) |
| Issues closed, last 90 days | 0 |
| Pull requests merged, last 90 days | 0 |
| Project age | 19 months |
Checked on 2026-10-09 with the GitHub API.
Project history
Daily GitHub snapshots since Oct 7, 2026. History covers stars and forks; health signals are not tracked historically.
Stars
492Forks
57Change since Nov 16, 2026: +1 stars·0 forks
Overview
mcp-remote-macos-use connects an MCP client, such as Claude Desktop, to a remote macOS desktop. It exposes screen capture and input controls so an AI agent can interact with Mac applications without installing a separate agent on the target Mac.
It is aimed at people who want an AI agent to carry out desktop workflows on a Mac they can access over Screen Sharing. The client-side server runs in Docker and uses configured Mac credentials and host details to connect.
Key Features
- Captures the remote Mac desktop for the AI client.
- Sends keyboard input and supports mouse movement, clicks, double-clicks, scrolling, and drag-and-drop.
- Opens or activates an application and returns its process ID.
- Scales mouse coordinates to the remote screen.
- Provides an MCP interface intended to work with MCP clients and different LLM providers.
- Documents optional WebRTC screen sharing through LiveKit.
- Runs as a Docker image, with no additional background agent required on the target Mac according to the project documentation.
Use Cases
- A Claude Desktop user can ask an agent to navigate Mac apps and perform multi-step desktop tasks.
- A researcher can automate work that involves reading and posting through desktop applications, as shown in the project demos.
- A video creator can have an agent interact with CapCut to assemble a short highlight video, as shown in a demo.
- A recruiting or marketing team can explore automating application workflows that involve Mail, LinkedIn, or Twitter, while checking actions and permissions carefully.
What you need
Detected in the repository
- Python >=3.10 (from pyproject.toml)
- A Dockerfile, so it can run in a container
- A test suite and automated checks on GitHub Actions
License in plain words
MITpermissive
- Commercial use: yes
- Modify and redistribute: yes
- You must keep: the copyright and license notice
- Share your changes: no
A summary, not legal advice: the LICENSE file is what applies.
Getting Started
Enable Screen Sharing on the target Mac, install Docker Desktop on the client Mac, and add the server to Claude Desktop's MCP configuration. The project publishes a Docker image; configure the target Mac credentials and hostname as environment variables:
{
"mcpServers": {
"remote-macos-use": {
"command": "docker",
"args": [
"run", "-i",
"-e", "MACOS_USERNAME=your_macos_username",
"-e", "MACOS_PASSWORD=your_macos_password",
"-e", "MACOS_HOST=your_macos_hostname_or_ip",
"--rm", "buryhuang/mcp-remote-macos-use:latest"
]
}
}
}
See the README for setup details and optional LiveKit configuration.
Alternatives
- MacOS-MCP: MacOS-MCP controls macOS apps locally through accessibility APIs and system tools rather than connecting to a remote Mac through Screen Sharing.
- open-codex-computer-use: open-codex-computer-use provides accessibility-based desktop automation across macOS, Linux, and Windows instead of remote Mac control through Screen Sharing.
| Project | Language | License | Stars | Status |
|---|---|---|---|---|
| mcp-remote-macos-use | Python | MIT | 490 | Inactive |
| MacOS-MCP | Python | MIT | 238 | Active |
| open-codex-computer-use | Swift | MIT | 2.3k | New |
Considerations
- Development is inactive, with no commits in over a year and no release in the past year. The project reports two contributors, with work concentrated heavily in one contributor.
- The server requires Python 3.10 or later according to its project configuration, and Docker for the documented deployment. The target Mac must have Screen Sharing enabled.
- The README says only Apple authentication protocol 30 is supported and describes it as using a 512-bit Diffie-Hellman prime. Treat this as a material compatibility and security concern; use only trusted, authorized connections and assess whether the protocol is acceptable for your environment.
- The README lists performance optimization and improved VNC cursor visibility as outstanding work. It also describes WebRTC through LiveKit, which requires a LiveKit server or LiveKit Cloud.
- The project is licensed under MIT. Review the license and your organization's obligations before adopting or redistributing it.
Found this useful?
Share it with someone who would like mcp-remote-macos-use.