Personal ยท 2026

System-Expert

Desktop system monitor with a built-in MCP server, so an AI agent sees exactly what you see.

GitHub
System-Expert screenshot

Problem

Monitoring tools put a separate backend service between the UI and system data, and AI agents have no structured, safe way to observe or act on a machine.

What it is

A native Tauri app that shows processes, CPU, memory, disk and network in real time. The same Rust binary hosts an MCP server with 22 tools, and a LangGraph agent on Groq answers questions in a built-in chat panel by reading and acting on the same data the UI renders.

Highlights

  • One Rust binary: the UI and MCP server share in-process data stores, with no backend to keep in sync
  • 22 MCP tools with paginated, context-window-sized results and a pid/name-guarded kill_process
  • Audit trail of every agent tool call, identifying the caller via the MCP handshake and the OS socket owner
  • Models swap mid-conversation with no restart; the app supervises the agent's lifecycle

Stack

  • Rust
  • Tauri
  • React
  • TypeScript
  • MCP (rmcp)
  • SQLite
  • WebSockets
  • Python
  • LangGraph
  • Groq