What is local-ai-setup?

This project is a step by step guide, not a packaged application, showing how to build a fully local AI browser automation assistant using four existing open source tools together: LibreChat as the chat interface, Ollama to run an AI model on your own machine, the Qwen3 8B language model, and Playwright MCP to let the AI control a Chrome browser. Because everything runs locally, no cloud AI service is required once it is set up. The setup works like a chain. You type an instruction in plain English into LibreChat, that request goes to the Qwen3 model running through Ollama, the model decides what browser actions are needed and sends them to the Playwright MCP server, which then actually opens Chrome, clicks, types, and navigates pages. The results of those actions are sent back so the AI can summarize what happened. The README walks through the whole process: installing Ollama, downloading the Qwen3 8B model, cloning and configuring LibreChat with Docker, starting the Playwright MCP server, installing Chrome for automation, and finally creating an AI agent inside LibreChat with recommended settings like temperature and reasoning effort. It also includes example prompts, such as asking the assistant to open a website and click a button, or to fill in a login form and confirm the result, plus a troubleshooting section for common problems like the MCP server not connecting or the model hallucinating. This is best understood as a recipe or reference setup rather than original software you install directly. It requires a reasonably capable computer, at least 16 gigabytes of memory, 20 gigabytes of free disk space, Docker Desktop, Node.js, and Git, with a GPU recommended but not required. There is no license file mentioned in the README, so it is unclear under what terms the instructions themselves may be reused.