Terminal UI engine in Go featuring AI agent driving, zero allocations, and 3D rendering.
Build testable, high-performance TUIs that AI agents can drive and tests can click.
Limoni is a terminal UI engine for Go designed for extreme performance and automation. It features a semantic tree that allows tests and AI agents to address widgets by name rather than screen coordinates.
The engine offers zero heap allocations per frame to eliminate garbage collection pauses, alongside native support for 3D software rasterisation, images, charts, and markdown. It provides both immediate mode and Elm architecture styles, plus built-in Model Context Protocol (MCP) support for AI agents.
Suits Go developers building demanding terminal applications, dashboards, or monitoring tools that require automated testing or AI integration. Not suitable if you require a stable 1.0 API today or rely heavily on the Charm ecosystem.
Lets Claude Code, Cursor, and other MCP clients drive your terminal application through dedicated automation tools.
Enforces zero heap allocations per frame for the renderer and common input widgets to prevent stutter.
Uses a semantic tree for Playwright-style UI tests that check wait conditions instead of sleeping.
Includes a 3D software rasterizer, image rendering, charts, and syntax-highlighted markdown out of the box.
Free under its open source license with no paid hosted options. The only cost is running and building it yourself.
880 open remote AI-training roles, updated November 15, 2026; most pay $20–$175 an hour. Zekai lists them free; the application goes to the employer’s platform.
Yes, Limoni includes limoni-mcp which provides MCP clients eight tools to interact directly with the semantic tree.
Yes, the engine compiles to WebAssembly allowing you to run your terminal applications directly in the browser.
Limoni enforces zero heap allocations per frame for its renderer, widget drawing, and common input handling to prevent garbage collection pauses.
See Zekai first in Google