SponsoredSkillry: Agent Skills with taste. Designed and tested in-house, so one prompt gives you polished slides, web pages and launch videos.Skillry: Agent Skills with taste Try free Skills
Agent Skills
  • Skills
  • Picks
  • ShopSoon
  • Solutions
  • Specification
  • Submit
Agent Skills
Agent Skills

Discover and share powerful Agent Skills for AI assistants

GitHubX (Twitter)
Skills
  • View All
  • Install Guide
  • Documentation
  • Solutions
  • Product Launch Video
  • Submit
Tools
  • Claude Code
  • Cursor
  • OpenCode
  • Codex CLI
  • Gemini CLI
Resources
  • anthropics/skills
  • awesome-claude-skills
  • vercel-labs/agent-skills
  • kepano/obsidian-skills
Company
  • About
  • Changelog
  • Privacy Policy
  • Terms of Service

Sponsored by Vectrust @ OpenDev Org & NextRouter Alliance

© 2026 EvergreenAI. Built with JimmyLv
x-algo-engagement - Agent Skill - Agent Skills
Home/Skills/x-algo-engagement

x-algo-engagement

6

Reference for X algorithm engagement types and signals. Use when analyzing engagement metrics, action predictions, or understanding what signals the algorithm tracks.

Download Folder
View on GitHub

CLI Install

Recommended

Use the skills CLI to install this skill with one command. Auto-detects all installed AI assistants.

Method 1 - skills CLI

npx skills i CloudAI-X/x-algo-skills/x-algo-engagement

Method 2 - openskills (supports sync & update)

npx openskills install CloudAI-X/x-algo-skills

Auto-detects Claude Code, Cursor, Codex CLI, Gemini CLI, and more. One install, works everywhere.

Installation Path

Download and extract to one of the following locations:

~/.claude/skills/x-algo-engagement/
SKILL.md

Skill Instructions

Back

Run with Cloud Agent

No setup needed. Let our cloud agents run this skill for you.

Select Provider

Select Model

ClaudeClaude Sonnet 5.5
$0.20/task

Best for coding tasks

Join Cloud Agent Waitlist

No setup required

X Algorithm Engagement Signals

The X recommendation algorithm tracks 18 engagement action types plus 1 continuous metric. These are predicted by the Phoenix ML model and used to calculate weighted scores.

PhoenixScores Struct

Defined in home-mixer/candidate_pipeline/candidate.rs:

pub struct PhoenixScores {
    // Positive engagement signals
    pub favorite_score: Option<f64>,
    pub reply_score: Option<f64>,
    pub retweet_score: Option<f64>,
    pub quote_score: Option<f64>,
    pub share_score: Option<f64>,
    pub share_via_dm_score: Option<f64>,
    pub share_via_copy_link_score: Option<f64>,
    pub follow_author_score: Option<f64>,
 
    // Engagement metrics
    pub photo_expand_score: Option<f64>,
    pub click_score: Option<f64>,
    pub profile_click_score: Option<f64>,
    pub vqv_score: Option<f64>,              // Video Quality View
    pub dwell_score: Option<f64>,
    pub quoted_click_score: Option<f64>,
 
    // Negative signals
    pub not_interested_score: Option<f64>,
    pub block_author_score: Option<f64>,
    pub mute_author_score: Option<f64>,
    pub report_score: Option<f64>,
 
    // Continuous actions
    pub dwell_time: Option<f64>,
}
Action Types by Category
Positive Engagement (High Value)
ActionProto NameDescription
FavoriteServerTweetFavUser likes the post
ReplyServerTweetReplyUser replies to the post
RetweetServerTweetRetweetUser reposts without comment
QuoteServerTweetQuoteUser reposts with their own comment
Sharing Actions
ActionProto NameDescription
ShareClientTweetShareGeneric share action
Share via DMClientTweetClickSendViaDirectMessageUser shares via direct message
Share via Copy LinkClientTweetShareViaCopyLinkUser copies link to share externally
Engagement Metrics
ActionProto NameDescription
Photo ExpandClientTweetPhotoExpandUser expands photo to view
ClickClientTweetClickUser clicks on the post
Profile ClickClientTweetClickProfileUser clicks author's profile
VQVClientTweetVideoQualityViewVideo Quality View - user watches video for meaningful duration
Negative Signals
ActionProto NameDescription
Not InterestedClientTweetNotInterestedInUser marks as not interested
Block AuthorClientTweetBlockAuthorUser blocks the author
Mute AuthorClientTweetMuteAuthorUser mutes the author
ReportClientTweetReportUser reports the post
Continuous Actions
ActionProto NameDescription
Dwell TimeDwellTimeContinuous value: seconds spent viewing post
How Scores Are Obtained

The PhoenixScorer (home-mixer/scorers/phoenix_scorer.rs) calls the Phoenix prediction service:

  1. Input: User history + candidate posts
  2. Output: Log probabilities for each action type per candidate
  3. Conversion: probability = exp(log_prob)
fn extract_phoenix_scores(&self, p: &ActionPredictions) -> PhoenixScores {
    PhoenixScores {
        favorite_score: p.get(ActionName::ServerTweetFav),
        reply_score: p.get(ActionName



Signal Interpretation
  • Scores are probabilities (0.0 to 1.0): P(user takes action | user sees post)
  • Higher = more likely: A favorite_score of 0.15 means 15% predicted chance of like
  • Negative signals have negative weights: High report_score reduces overall ranking
  • VQV requires minimum video duration: Only applies to videos > MIN_VIDEO_DURATION_MS
Related Skills
  • /x-algo-scoring - How these signals are combined into a weighted score
  • /x-algo-ml - How Phoenix model predicts these probabilities
Follow Author
ClientTweetFollowAuthor
User follows the post's author
Dwell
ClientTweetRecapDwelled
User dwells (pauses) on the post
Quoted ClickClientQuotedTweetClickUser clicks on a quoted post
::
ServerTweetReply
),
retweet_score: p.get(ActionName::ServerTweetRetweet),
// ... maps each action to its probability
}
}