cclank

cclank / x-algorithm-wiki

Public

Architecture wiki for the open-sourced X "For You" recommendation algorithm (xai-org/x-algorithm) — 21 source-anchored pages

lansu-wiki-web.lank.workers.devwikicclankx-algorithm-wiki#index
83
7
89% credibility
Found May 18, 2026 at 88 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
AI Summary

This is an educational knowledge base that explains how X (formerly Twitter)'s 'For You' recommendation algorithm works. Based on open-source code released by xAI, the wiki contains 34 interconnected pages: 11 beginner-friendly guides written in plain language, plus 23 technical pages with source code references. The project helps regular people understand how social media feeds decide what content to show them—covering topics like how posts get selected, how machine learning ranks content, how ads are mixed in, and common misconceptions about visibility and shadowbans. It's designed to be browsed like a wiki or imported into Obsidian for personal knowledge management.

How It Works

1
💬 You hear X made their algorithm public

You learn that X (Twitter) released the code behind your 'For You' feed, and you're curious how it decides what to show you.

2
📚 You find a friendly guide that explains it

You discover this wiki with 11 plain-language pages that promise to break down the algorithm without requiring any coding knowledge.

3
You choose your learning path
⚡
Quick overview first

Start with the 'how it works' page to see the whole system at a glance

🎯
Dive into specifics

Jump to topics like 'how posts are picked' or 'visibility' that answer your specific questions

4
💡 The pieces finally click together

You read how five components work together—how posts are found, scored, ranked, filtered, and mixed with ads—to create your feed.

5
🔍 You verify claims against real code

Every explanation includes actual code snippets and line numbers, so you can trust the information is accurate.

🎉 You understand your feed

You now know exactly how X's algorithm works—from the moment someone posts to when it appears (or doesn't) in your For You feed.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 88 to 83 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is x-algorithm-wiki?

This is a documentation project that reverse-engineers X's (Twitter's) "For You" recommendation algorithm from the open-source xai-org/x-algorithm repository. It translates 6,800+ lines of Rust and Python code into 34 interconnected wiki pages that explain how posts get selected, ranked, and delivered to users. The docs are split into two tracks: 11 plain-language guides for quick understanding, and 23 technical pages with real code snippets and source anchors. You can browse it online or import it into Obsidian for local navigation.

Why is it gaining traction?

The hook is credibility through traceability. Every claim links back to specific source lines, so you're not reading speculation -- you're reading analyzed code. The "operating myths vs. source truth" page alone is worth it for anyone who's spent time on algorithm conspiracy theories. The dual-track structure means you can skim the conceptual guides first, then drill into technical details only when needed. It's also one of the few resources that maps the full pipeline end-to-end, from Kafka ingestion through Grok transformer ranking to ad blending.

Who should use this?

ML engineers building recommendation systems will find the pipeline framework and scoring/ranking documentation directly useful. Product managers and growth engineers curious about shadowban mechanics, cold start behavior, or engagement manipulation will get value from the plain-language guides. Researchers studying social media algorithms have a rare open-source reference point. Anyone else will at least satisfy their curiosity about why that one tweet went viral.

Verdict

This is a genuinely useful reference for anyone working with or around recommendation systems, backed by actual source code analysis. The credibility score of 0.9% reflects its niche focus and early stage -- 83 stars is modest, and it's a documentation-only project with no runtime component to test. That said, the depth and cross-linking are solid. Bookmark it if you're evaluating x-algorithm or just want to understand how modern feeds work under the hood.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.