Drop your tweet, thread, or screenshot. We score it across 13 engagement signals, predict how it'll perform, and rewrite it stronger in under 30 seconds.
🚨 YouTube content creators, take note! How to redub any YouTube video into 100+ languages in <2 minutes (for free):1Hook written • Auto-download & audio separation • AI transcription & translation • Voice cloning & seamless redubbing2Format improved And all that in **under 2 minutes**, 100% local and completely free. Which language should I test this on next? 👇3Reply trigger
Every draft is scored against the engagement actions X's open-source ranker explicitly tries to predict. Ten the algorithm rewards. Four it punishes.
xAudit reads the same engagement signals from X's open-source ranker, so every recommendation is verifiable.
Recommendations are tied to the prediction labels the open-source ranker optimizes for.
xai-org/x-algorithmDrop your draft, thread or a screenshot of someone else's tweet you want to learn from. Multimodal in.
We score it across all 13 signals — 10 the ranker rewards, 4 it punishes — with the line that triggered each judgment.
Three rewrites, each strengthening a different weak signal. Copy the one that fits the angle you want.
Tools in this category overpromise. We'd rather underclaim, source everything, and let you decide.
No. Drafts are sent to the grader and discarded after the response. We don't have an account system to store anything against. If we add accounts, storing drafts will be opt-in.
Those tools optimize against folklore: post-time charts, hashtag counts, link-in-comments tricks — claims that aren't in X's open code. We only grade against the engagement signals X published themselves in xai-org/x-algorithm.
No, and anyone who promises that is lying. We grade the signals the ranker tries to predict — that's the direction. Audience fit, timing and luck are still real. We surface what you can control.
Because we only use the signals documented in X's open-source ranking repository at github.com/xai-org/x-algorithm. We do not have the numeric weights — nobody outside X does. We work on direction (which signals matter), not magnitude.
They're proprietary. We never guess at them. That's why every grade is a band (Weak/Moderate/Strong) instead of a percentage — we refuse to fake precision.
Not yet. xAudit is built on X's repo specifically. We'd rather be excellent on one platform than guess on five.