Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
What This Skill Does
Scores and optimizes tweet drafts using rules derived from X's open-source ranking algorithm. X's For You feed ranks tweets via a transformer model (Phoenix) that predicts 19 engagement actions; this skill encodes those structural rules into a scoring engine. Works for single tweets and threads.
Scores against X's actual open-source ranking weights rather than generic social media tips.
When to use it
- Scoring a draft before posting to estimate For You reach
- Rewriting a tweet that got less engagement than expected
- Structuring a thread for maximum first-tweet hook performance
- Diagnosing why a specific past tweet underperformed
- Deciding how many tweets to post in a day without triggering author diversity penalties
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: Score this tweet: 'Just launched our product! Check it out at example.com #startup #launch'
- 1Read references/algorithm-rules.md for the complete rules engine
- 2Analyze the draft across all weighted categories: reply potential, media, shareability, dwell time, content quality, format, and negative signals
- 3Flag specific issues: link in body incurs learned penalty, two hashtags reduce engagement score, no question or CTA lowers P(reply)
- 4Calculate weighted score and assess each of the 19 predicted engagement actions
- 5Output score card with per-category indicators, predicted action boosts, and a rewritten optimized version
Score card showing ~32/100 with suggestions to move the link to a reply tweet, remove hashtags, add a question CTA, and a rewritten version under 180 characters