Skip to content

mcp-tuikit

mcp-tuikit is a powerful Model Context Protocol (MCP) server that enables AI agents to orchestrate, interact with, and monitor Text User Interfaces (TUIs) and headless terminal applications.

🚀 Fully Cross-OS Compatible
Designed from the ground up to run seamlessly on macOS, Linux, and Windows. It bridges the gap between LLMs and native terminal environments, allowing agents to "see" (via PNGs and text) and "type" (via keystrokes) across any operating system.

What Does It Do?

  • Spawn & Orchestrate: Launch complex terminal applications (nvim, htop, git, or standard shells) using backends like tmux, xterm.js, or native emulators (Alacritty, WezTerm, iTerm2, etc.).
  • Capture State: Take accurate textual representations and pixel-perfect visual PNG screenshots of running terminal apps.
  • Headless Execution: Run terminal sessions invisibly in CI environments (via Xvfb, Sway, or kwin on Linux, or xterm.js everywhere) without interrupting the user's workflow.
  • Flow Engine: Execute pre-defined, step-by-step TUI interaction scripts (YAML-based) for testing or autonomous agent tasks.

Why mcp-tuikit?

Traditionally, giving an AI agent access to a terminal means piping raw stdout/stderr. This fails spectacularly for curses-based applications (like vim or fzf) that rely on complex ANSI escape sequences and screen redrawing.

mcp-tuikit solves this by running an actual PTY (pseudo-terminal) backed by tmux and rendering it to the agent either as a clean text grid or a visual screenshot. The agent sees exactly what a human would see.

Next Steps