硬件系统基准测试工具

🌌 mcp-sysinfo 🌌

A beautiful, AI-powered hardware & system benchmarking tool for your local machine

Welcome to mcp-sysinfo! 🌟 This project is a shiny hardware & system benchmarking tool that allows an Artificial Intelligence to run tests and analyze your computer's performance in real time!

If you are a new Python programmer, this is a fantastic example of how to connect a large language model (like Gemini) to actual Python code that runs locally on your machine using the magic of the Model Context Protocol (MCP).

🧐 What is this project doing?

Normally, an AI like ChatGPT or Gemini is stuck in the cloud ☁️. It doesn't know how much RAM your computer has, and it can't run a speed test on your CPU.

This project bridges that gap by giving the AI "tools" 🛠️ it can use!

Here is how it works step-by-step:

1. 🗣️ You ask a question: You type something like, "Is my CPU fast?"
2. 🧠 The AI receives the question: The AI realizes it needs to run a CPU benchmark tool to answer your question.
3. 📡 The AI asks for data: The AI sends a message saying, "Hey, please run the tool called run_cpu_benchmark for me."
4. 💻 The Python Server obeys: Our local Python script actually runs the heavy math calculations on your computer to test the CPU, and gets a score.
5. 📤 The Python Server replies: The script sends the score back to the AI.
6. ✨ The AI gives you the answer: The AI reads the score and writes a human-friendly response like, "Your CPU scored 300,000, which is very fast!"

🏗️ Why use MCP? (The Architecture)

We use the Model Context Protocol (MCP) to keep things perfectly organized and totally secure 🔒. We split the code into two main parts:

1. ⚙️ The Server (mcp_server.py): This script contains the actual Python benchmarking functions (like measuring IO speeds). It acts as a menu, saying "Here are the 4 tools I know how to run."
2. 🤖 The Client (mcp_client.py): This script talks to the Gemini API. It shows the AI the "menu" of tools from the server. When the AI wants to use a tool, the client passes the message to the server, waits for the result, and gives it back to the AI.

💡 Security Note: This separation means the AI is safely restricted. It can't run arbitrary commands on your computer; it can only run the specific benchmarking tools you provided in the server script!

📂 Deliverables inside this folder

• 📜 mcp_server.py: The MCP Server. Exposes system tools like run_cpu_benchmark().
• 🕹️ mcp_client.py: The MCP Client. Instantiates Gemini, fetches the tools, asks the model to benchmark the system, fields its tool usages, and prints the AI's final system analysis.
• 📦 sample_benchmark.json: An example payload of what the underlying MCP raw tool outputs look like when aggregated.

🛠️ Prerequisites

1. Install required dependencies:
   复制

   pip install google-generativeai flask python-dotenv
   

2. Create a .env file in the root directory and add your key:
   复制

   GEMINI_API_KEY="your-gemini-api-key"
   

🎮 Example Run

🌐 Web UI (Tailwind CSS)

You can run a beautiful web interface to interact with the benchmarking assistant!

# Ensure you are using the virtual environment if applicable
python3 app.py

Then, open http://127.0.0.1:5001/ in your browser. 🌍

Features:

• 💬 Type a custom prompt in the chatbox, e.g., "Just check my memory and IO speeds."
• 🚀 Click Ask to let Gemini run the corresponding tools and stream the insights back!

💻 CLI Mode

You can also run the client directly from the command line, and optionally provide a custom prompt:

python3 mcp_client.py --prompt "Only run the CPU benchmark and tell me if it's fast enough."

📸 Example Prompts

Here are a few different demo command prompts you can try to see how the GenAI controller uses its available tools in real time!

🎯 Prompt triggering a single API/tool:

• "Only run a CPU benchmark and tell me if its score is good." -> Triggers run_cpu_benchmark()
• "What is my system info?" -> Triggers get_system_info()

🤹‍♀️ Prompt triggering multiple APIs/tools sequentially:

• "Check my memory and IO speeds, then summarize the results." -> Triggers run_memory_benchmark() and run_io_benchmark()
• "Run all performance diagnostics to give me a hardware analysis." -> Triggers get_system_info(), run_cpu_benchmark(), run_memory_benchmark(), and run_io_benchmark()

🔄 Flow Outcome Recap

1. 🤝 Handshake establishes JSON-RPC connection.
2. 🔍 Tools are discovered and fed to Gemini's schema parameters.
3. 🤔 Gemini decides to use get_system_info, run_cpu_benchmark, run_memory_benchmark, and run_io_benchmark.
4. 🏃‍♂️ The isolated MCP server executes these, returning metrics like CPU scores, memory operations/sec, and disk read/write bandwidth in JSON.
5. 📊 Gemini digests the JSON structured data and summarizes bottlenecks and insights into a friendly AI Analyst Report!