You post on X; some days people read you, other days nobody does, and everything said about the algorithm is rumor and supposed hacks. This repo is the code X published for deciding what goes into the For You feed: where posts come from, which filters apply and how the score that ranks them is calculated.
Who this is NOT for
If your customers are not on X because you sell through WhatsApp, Instagram or TikTok, skip it: this code covers only X's For You feed and nothing about other networks. It also won't help if you want to build a recommendation system for your store or app, because it is built for X's infrastructure and the README itself warns that pieces are missing. And if you are looking for a "post at this hour" trick, it is not here.
The value here is not the code, it is the myths it debunks. The README corrects the most common mistake: the weights on Like, Share or Report multiply the predicted probability that THAT viewer takes the action, not raw counts. Reading "1 report wipes out 468 likes" is an error the repo itself points out. In practice, according to its diagram, posts older than 48 hours are dropped before ranking, several posts in a row from the same author lose score, content from accounts the viewer does not follow gets a discount and new authors get a boost. In short: your real followers outweigh any hack. The repo is active (last push October 3, 2026), and since September 18 its Under the Hood tool lets you see whether your account or posts carry labels that limit their reach.
Your first step
Open the repo's README on GitHub and read only the "August 14th, 2026" section on how weights work, then the step 4 "PRE-SCORING FILTERS" diagram. In 15 minutes you will know what keeps one of your posts from even entering the race.
The X For You Feed Algorithm implementation guide
The project belongs to SpaceXAI Org (xai-org) and it is free: the link beside asks for nothing. The guide to set it up without getting lost is ours: step by step, common mistakes and a checklist. Leave your email and we will send it.
No spam. Unsubscribe anytime.