allcanuse-mcp

An MCP Server for Windows / Linux experimental environments, designed to expose common local machine capabilities to local Agents / MCP Clients.

License: MIT

The goal of allcanuse-mcp is not just to provide a few scattered tools to the model, but to systematically expose "what this experimental machine can do" to the model. Once connected, the model can not only read files and run commands, but also use the local machine as an actionable workstation.

Currently, 90+ tools are provided, with core capabilities including:

• System and Environment Probing: Read system architecture, time, disk space, environment variables, network configuration, network adapters, current IP, and basic runtime environment information.

• Command Execution and Automation: Execute cross-platform shell, Windows cmd, and PowerShell, with support for using command output for subsequent decision-making.

• Process and Port Management: Start processes, terminate processes, list processes, view process trees, locate processes by port, and enumerate listening ports and established connections.

• File and Code Editing: Directory tree traversal, search by filename, search by text, read long files in chunks, precise line replacement, fixed text replacement, text file writing, JSON read/write, binary read/write, hashing, recent files, and desktop file viewing.

• Development Assistance and Engineering Operations: Compression/archiving, extraction, configuration modification, source code localization, local patching, and running verification commands, suitable for letting the model directly fix bugs, modify configurations, write scripts, and organize projects.

• Web and HTTP Operations: HTTP requests, HEAD probing, response header retrieval, web content extraction, web to Markdown, link extraction, table extraction, specific HTML element extraction, web form submission, file upload, file download, and redirect chain tracking.

• Network Diagnostics and Protocol Debugging: DNS resolution, reverse resolution, TLS certificate reading, ping, traceroute, TCP connectivity testing, raw TCP send/receive, UDP send/receive, WebSocket debugging, and small-scale port scanning.

• Desktop and Window Observation: Enumerate windows, get foreground window, summarize desktop context, and take desktop screenshots, suitable for working with models that have visual capabilities.

• Camera Capabilities: Enumerate local cameras and take photos, suitable for experimental scenarios requiring visual input.

• Duty and Long-running Task Hosting: Wait for time, wait for file/process/port/HTTP/window/desktop changes, background duty tasks, task scheduling, task events, task artifact recording, and handover summaries after disconnection.

• Model Self-inspection and Self-discovery: Can directly list all tools and complete descriptions, making it easy for the model to decide what to call next in unfamiliar tasks.

What can the model do after connecting?

After connecting to this MCP, the model can do more than just "answer questions"; it can actively advance tasks on the local machine, for example:

• Help you troubleshoot "why a service won't start": Check ports, then processes, then logs, then send HTTP requests, and finally provide the root cause.

• Help you modify code and fix configurations: Search for target files and functions, read large files in chunks, precisely modify a few lines of code, and then automatically run verification commands.

• Help you read web pages and organize information: Scrape web content, extract links, extract tables, download attachments, and summarize the results into a structured format suitable for further processing.

• Help you check local network issues: Check DNS, ping, ports, TLS, routing, and connection status to determine if it's a resolution issue, host unreachable, port blocked, or an application-layer exception.

• Help you automatically observe desktop changes: Monitor installers, window pop-ups, foreground switching, and automatic screenshots, suitable for GUI tasks in experimental environments.

• Help you upload and download local files: Upload logs, archives, and build artifacts to interfaces, or pull network files to the local machine like a lightweight wget.

• Help you perform long-term duty: Wait for files to be generated, services to recover, or windows to appear while the user is away, and quickly take over via a handover summary upon return.

• Help you automate experiments in virtual machines: Hand over the entire "observe -> judge -> execute -> verify" process to the model, rather than letting the model stop at verbal suggestions.

If you want the model to be more like a truly hands-on local assistant, rather than one that only talks but doesn't act, allcanuse-mcp is positioned to put these capabilities directly into the model's hands.

Related MCP server: windows-admin-mcp

Quick Start

Run directly in the repository root:

python run_server.py --transport stdio

Windows is also supported:

start.cmd --transport stdio

Linux is also supported:

./start.sh --transport stdio

If already installed as a package, you can also:

pip install -e .
allcanuse-mcp

Documentation

• General Usage Instructions: docs/USAGE.zh-CN.md

• Overview of All Tools: docs/TOOLS.zh-CN.md

• Duty Workflow Explanation: resource://guides/workflows/duty-watch

• LM Studio Integration Tutorial: docs/LM-STUDIO.zh-CN.md

• Codex / Claude Code / OpenCode Integration Tutorial: docs/CLIENT-INTEGRATIONS.zh-CN.md

• End-user Usage Tutorial after Release: docs/RELEASE-USAGE.zh-CN.md

If you want to connect this MCP to different clients, it is recommended to prioritize reading:

• LM Studio

• List of All Tools

• ChatGPT Codex / Codex CLI

• Claude Code

• OpenCode

• Installation and Integration for End Users after Release

Optional Dependencies

• Camera functionality depends on opencv-python

• Linux window enumeration depends on wmctrl

• Linux active window query depends on xprop

• Linux screenshots may use gnome-screenshot, scrot, or imagemagick

Verification

python -m compileall src tests run_server.py
$env:PYTHONPATH=(Resolve-Path .\src).Path
python -m unittest discover -s tests -v