Analysis updated 2026-05-18
Give an AI coding agent grounded knowledge of X's ranking algorithm instead of guesses
Ask an agent to explain how X ranks posts in the For You feed with citations to source code
Audit or design a recommendation system using X's documented architectural choices
| kkkhushman/x-algo-skill | 0verflowme/alarm-clock | 0xhassaan/nn-from-scratch | |
|---|---|---|---|
| Stars | 0 | — | 0 |
| Language | — | CSS | Python |
| Last pushed | — | 2022-10-03 | — |
| Maintenance | — | Dormant | — |
| Setup difficulty | easy | easy | moderate |
| Complexity | 1/5 | 2/5 | 4/5 |
| Audience | developer | vibe coder | developer |
Figures from each repo's GitHub metadata at analysis time.
This is a knowledge skill, a packaged reference document that AI coding agents can install to gain deep, accurate understanding of X's open-sourced For You feed algorithm, released in May 2026 from the xai-org/x-algorithm repository. A "skill" in this context is a collection of structured reference files that an AI agent loads into its context so it can reason correctly about a topic. Instead of an agent guessing how X's feed works based on potentially outdated training data, it loads this skill and gets citation-grounded information tied to specific files and lines in the upstream source code. Once installed, an agent can explain how X ranks posts in the For You feed, help design or audit a recommendation system based on X's architectural choices, answer questions about specific components (Phoenix model, Thunder, Grox, home-mixer, the candidate pipeline, and ads blending), and flag the 22 documented ways agents typically reason incorrectly about the algorithm. The skill is installed via the skills.sh command-line tool using a single command, and it works across more than 50 AI coding tools including Claude Code, Cursor, Codex, Copilot, Gemini CLI, and others. The format follows the Agent Skills specification, a SKILL.md entrypoint with YAML metadata, plus supporting markdown files organized by topic: architecture overview, the Phoenix ranking model, request lifecycle, ranking weights, content filtering, common pitfalls, and a glossary of internal terminology. The skill was built against the May 15, 2026 release of xai-org/x-algorithm and is licensed under Apache-2.0, matching the upstream source.
A reference skill that AI coding agents install to accurately understand and reason about X's open-sourced For You feed ranking algorithm.
Use freely for any purpose, including commercial use, as long as you keep the copyright notice and state changes.
Setup difficulty is rated easy, with roughly 5min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.