LumaBrowser

•
154
•

Unify local AI model execution, multimodal media generation, and agentic browser automation within a privacy-focused Chromium desktop application.

LumaBrowser integrates a complete Chromium web browser with local AI runtimes like llama.cpp and MLX to run language, image, video, and voice models entirely on your own hardware. Designed to eliminate fragmented AI toolchains, it includes a built-in Model Context Protocol (MCP) server alongside CDP and WebDriver support, allowing external agents such as Claude Code, Codex, and Cline to interact with live, authenticated browser tabs. By keeping model inference and computer-use automation completely local, it provides a private, zero-token-cost environment for browsing, coding, and workflow automation.

Key Features

01Built-in MCP server enabling AI coding assistants like Claude Code and Cline to navigate and drive real browser tabs
02Hardware-aware local model runtimes supporting llama.cpp and MLX across NVIDIA, AMD, Intel, Qualcomm, and Apple Silicon GPUs
03Full-featured Chromium browser complete with Chrome extension support, ad blocking, and isolated agent workspaces
04Offline multimodal capabilities including local image generation, video, text-to-speech, and voice cloning
05OpenAI- and Anthropic-compatible local endpoints alongside WebDriver and CDP for custom script and agent integration
06154 GitHub stars

Use Cases

01Empowering AI coding assistants to automate tasks, scrape data, and test web apps using existing authenticated browser sessions
02Running privacy-sensitive LLM, vision, and speech inference locally without incurring cloud API costs or leaking data
03Automating complex, repetitive web browsing workflows and desktop tasks using local, trigger-based AI agents