wtf is ember?

ember is a lightweight tool that lets an AI agent browse the web, similar to what a person's web browser does, but built to be small and simple to set up. Most existing options for giving an AI access to the web either bundle a full web browser, which can be a few hundred megabytes, or require Docker to run. ember instead sits idle at around 17 megabytes and only reaches for browser style rendering when a page actually needs it, otherwise it just fetches the page content directly. It is installed with a single pip command, no Docker and no API key required to get started, and can be used three ways: as a command you type in a terminal, as a Python library you import into your own code, or as a small local web server you can send requests to. From the command line you can fetch a single page as readable text, search the web, crawl an entire site up to a page limit, list every URL a site contains, or point it at a page with a plain English instruction like finding a product under a certain price. Some of these features, specifically the ones that ask an AI model to read a page and follow instructions or pull structured data out of it, need you to supply your own API key for a language model. Results can be saved automatically to a folder, or to a specific file you name, and the tool supports fetching many web pages at once instead of one at a time. Windows is supported for most features, though the full browser rendering capability specifically needs Linux or a Linux compatibility layer on Windows. The project is open source and released under the MIT license, and is aimed at people building AI agents that need to look things up on the web without adding a heavy browser dependency to their setup.