X is expanding its open-source codebase to include more of the systems behind its default For You timeline, as the social network seeks to give developers, researchers and users more insight into how posts are ranked.
The company said on Thursday that it will make the code for the For You feed available on GitHub under the Apache v2 licence. The release includes the core system that retrieves posts, ranks them for individual users and assembles the timeline.
X said the expanded codebase will also provide more detail on model configuration, filters and core ranking systems. It will include parameters that weight different signals, helping to show how the platform determines which posts users see.
The company said the codebase will be around 10 to 15 times larger than its earlier open-source release.
Keith Coleman, X’s vice-president of product, said developers would be able to inspect the ranking system, some filtering tools and the scores calculated for posts.
“You’ll get the core ranking code that pulls posts and ranks them for any given user and assembles the feed,” Coleman told TechCrunch.
He said some systems, including elements of the ranker and scoring tools, could be run outside the company.
A new ‘Under the Hood’ tool
Alongside the code release, X is introducing a transparency tool called Under the Hood.
The tool will allow eligible users to download aggregate data on how X’s ranking systems have affected their accounts and posts. Users who have posted at least 10 times in the previous month will be able to download a JSON file showing whether labels were applied to their account or posts during the preceding calendar month.
The company said the feature will initially be available to a test group of accounts that are at least one year old before a wider rollout.
X said users without technical expertise could upload the file to a large language model and ask it to interpret the information alongside the company’s GitHub repository.
The move could give users more visibility into whether their posts have been affected by systems that shape distribution across the platform.
What X is not releasing
Some systems will remain private, including tools that use Grok to predict whether a post could violate X’s rules.
The company said those systems were excluded to prevent bad actors from using the information to evade moderation and flood the platform with spam.
X said it had already shared the expanded codebase with external researchers familiar with recommendation systems. According to Coleman, some researchers were able to run the numerical scoring system used for posts outside the company.
Developers will also be able to submit proposed changes through GitHub pull requests. X engineers will decide whether to incorporate those submissions into the platform’s algorithm.
A transparency test
The announcement comes amid continuing concerns over how social-media ranking systems influence political debate, elections, misinformation and public discourse.
X said its goal was to allow the public to assess how posts are distributed, examine whether the system operates fairly and offer criticism that could improve it.
The company has made transparency a central part of its argument for opening the code. But its wider disclosures have become more limited since it returned to private ownership under Elon Musk, with less public reporting on areas such as user metrics, revenue and government takedown requests.
The code release and new transparency tool offer a more detailed view of the mechanics behind X’s feed. Their significance will depend on how accessible the information proves to users, how much of the platform’s ranking process remains visible and whether the company acts on external criticism.
