Hi everyone,
I want to share a free open-source project I built for people using AI in their projects, research, coding, documentation, or automation workflows.
AI often has to process a lot of unnecessary content before reaching the useful part of a page: menus, cookie banners, repeated links, comments, navigation, and other noise.
That extra context costs tokens, increases waiting time, and can make the final answer less focused.
aidigest helps clean that information before it reaches your AI. It gives the model a shorter and more useful version of the page, while showing the before-and-after token count so you can see whether it actually helped.
It is free, local-first, and does not require a subscription or API key to try. Windows users can simply download one portable EXE. It can be used with Claude Desktop, Cursor, OpenCode, VS Code, and other AI workflows.
One included example goes from 259 tokens to 126 tokens — around 51% less. Results depend on the page, so there is no fixed saving promise.
There is no token sale, investment scheme, or hidden paywall. It is simply free software designed to help AI projects process less unnecessary data.
You can try it here:
https://github.com/iarttools/aidigestIf you work with large documents, websites, GitHub repositories, support pages, or research material, I would really appreciate your feedback.
