mcp-my-mac
The mcp-my-mac server provides real-time Mac system and Conda environment information for AI assistants like Claude to offer targeted assistance and troubleshooting.
This server allows:
Access to Mac System Information: Retrieve detailed hardware, software, and configuration data via
system_profiler(WiFi, network, power, storage, etc.)Conda Environment Analysis: Get comprehensive details about Conda installations, environments, packages, and Python setups
GPU Availability Checks: Assess Apple Metal (MPS) GPU acceleration for PyTorch/TensorFlow, including performance benchmarks
AI-Driven Optimization: Enable system-specific insights for software optimization and issue resolution
The server runs as a secure local API, executing only verified commands with minimal overhead.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@mcp-my-mac what Python environments do I have installed?"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
MCP My Mac
A lightweight server that exposes Mac system information via a simple API, allowing AI assistants like Claude to access real-time system information about your Mac. This tool is primarily designed for Mac users who want to experiment with AI and Deep Learning on their machines.
Status: BETA - This project is currently in beta. We're actively looking for feedback to improve functionality and user experience. Please share your thoughts and suggestions!
Why Use It?
Provides Claude Desktop or other MCP clients with access to your Mac's hardware specifications, system configuration, and resource usage
Enables more targeted and accurate assistance for software optimization and troubleshooting
Runs as a secure local API with minimal overhead
Only executes safe, verified commands:
system_profiler- to gather system informationconda- to analyze Python environment configurations
Related MCP server: Software Management MCP Server
Installation
Method 1: Using UV + Git Clone
Prerequisites
Python 3.8 or higher
UV package manager installed
Steps
Clone the repository:
bash git clone git@github.com:zhongmingyuan/mcp-my-mac.gitConfigure for your AI client:
[Claude Desktop] Add the following to your MCP server config file:
"mcpServers": { "mcp-my-mac": { "command": "uv", "args": [ "--directory", "/YOUR_PATH_TO/mcp-my-mac", "run", "-m", "mcp_server_my_mac" ] } }Note: Replace
/YOUR_PATH_TOwith the actual path where you cloned the repository.[Cursor] Add tool by selecting "command" in UI:
uv run --directory /YOUR_PATH_TO/mcp-my-mac mcp_server_my_mac
Usage
After installation, Claude Desktop will automatically connect to this API when running on your Mac, allowing it to access system information when needed for answering your questions or providing assistance.
Available Tools
3 toolsmcp_call_conda_infoA
Get comprehensive information about the Conda installation on this system.
If env_name is provided, it will return the information for the specified
environment as well.
Returns detailed information including:
- Conda version and configuration
- Python version and virtual packages
- Base environment location
- Channel URLs and package cache locations
- Platform and system details
- Complete list of all Conda environments with their paths
- Complete list of all packages in the specified environment and their versions
This is useful for diagnosing Conda-related issues or understanding
the Python environment configuration on this system.
Inputs: env_name. Returns: result
mcp_call_gpu_availableA
Check if GPU is available in torch for a specific conda environment.
Input: torch or tensorflow
if framework is not provided, it will default to torch.
Returns a detailed dictionary with the following information:
- "torch_version": PyTorch version string
- "python_version": Python version string
- "platform": Platform information string
- "processor": Processor type
- "architecture": CPU architecture
- "mps_available": True if MPS (Metal Performance Shaders) is available
- "mps_built": True if PyTorch was built with MPS support
- "mps_functional": True if MPS is functional, False otherwise
- "benchmarks": A list of benchmark results for different matrix sizes, each containing:
- "size": Matrix size used for benchmark
- "cpu_time": Time taken on CPU (seconds)
- "mps_time": Time taken on MPS (seconds)
- "speedup": Ratio of CPU time to MPS time (higher means MPS is faster)
This helps determine if GPU acceleration via Apple's Metal is properly configured
and functioning, with performance benchmarks for comparison.
Inputs: env_name (required), framework.
mcp_call_mac_system_profilerA
Call the system_profiler with the given datatype. Allow LLM to deepdive into the
system information.
This function is used to get the system information to help user to understand the
system and potentially debug.
Allowed datatypes:
- SPAirPortDataType - Airport/WiFi information
- SPApplicationsDataType - Application information
- SPAudioDataType - Audio device information
- SPBluetoothDataType - Bluetooth information
- SPCameraDataType - Camera information
- SPDiagnosticsDataType - Diagnostic information
- SPDisplaysDataType - Display and graphics information
- SPFirewallDataType - Firewall settings
- SPHardwareDataType - Hardware specifications
- SPLocationDataType - Location services information
- SPMemoryDataType - Memory information
- SPNetworkDataType - Network settings and interfaces
- SPNVMeDataType - NVMe storage details
- SPPCIDataType - PCI devices information
- SPPowerDataType - Battery and power information
- SPSoftwareDataType - Software and OS information
- SPStorageDataType - Storage devices and volumes
- SPThunderboltDataType - Thunderbolt ports and connections
- SPUSBDataType - USB devices and connections
Inputs: datatype (required). Returns: result
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 3 tools
Each tool has a clearly distinct purpose with no overlap: conda_info focuses on Conda environments, gpu_available checks GPU availability and performance, and mac_system_profiler retrieves system information. The descriptions clearly differentiate their domains, making misselection unlikely.
All tools follow a consistent 'mcp_call_' prefix with descriptive snake_case suffixes (conda_info, gpu_available, mac_system_profiler). This uniform pattern makes the tool set predictable and easy to understand.
With only 3 tools, the server feels thin for a 'my-mac' system management scope. While the tools cover Conda, GPU, and system profiling, there are likely other Mac/system-related operations missing, making the set feel incomplete rather than well-scoped.
For a Mac system management server, there are significant gaps: no tools for file operations, process management, network utilities, or basic system commands. The three tools provide deep but narrow coverage, leaving many common system tasks unaddressed and likely causing agent failures for broader requests.
Related MCP Connectors
Agent Replay Debugger MCP — record every agent step + deterministic replay. Step-debugger for
Check packages for CVEs, slopsquatting, and CISA KEV before your AI agent installs them.
Real-time Python package and vulnerability data for AI coding agents.
Package intelligence for AI agents across npm, PyPI, crates.io and deps.dev. No API keys.
Related MCP Servers
- AlicenseBqualityFmaintenanceProvides up-to-date Python package manager commands by cross-referencing official pip, poetry, uv, and conda documentation with automatic weekly updates.140MIT
- FlicenseNot gradedqualityDmaintenanceEnables AI agents to automate local software management workflows including installation, uninstallation, updates, and environment recommendations. It provides a standardized, non-interactive interface for managing system applications and tracking software versions safely.-
- AlicenseCqualityCmaintenanceProfessional local context management and system diagnostic tools for AI IDEs (Cursor, Trae, Antigravity, Windsurf).3684 npm1MIT
- AlicenseNot gradedqualityCmaintenanceVet a package before your AI coding agent uses it — authoritative facts (CVEs, license, maintenance) via an MCP server + CLI. Local, no account.102 npm13Business Source 1.1