360 Secure SkillHub
360 安全技能中心

X Search Highlights

通过浏览器自动化脚本在 X 搜索指定主题，滚动加载候选帖子，按点赞、转发、浏览量的加权互动分排序并过滤，可按 Markdown 或 JSON 输出高价值帖文列表；依赖 bb-browser 与已登录的 X 会话。

ghostwritten
ghostwritten
热度 41.77
最近更新 2026-05-11 17:40:23
分类：搜索依赖：联网浏览搜索 / 代码执行应用场景：社交媒体 / 研究 / 摘要总结 / 工作流

安装技能

复制下面的话术发给你的 Agent，它会通过 secure-skills 检索并安装该技能。

执行命令安装此 skill： npx secure-skills-cli add x-search-highlights
Metadata
name
x-search-highlights
description
Search and extract high-value posts from X (Twitter) with engagement-based ranking. Use when user asks to search X, find best posts, get highlights from Twitter/X, or extract high-engagement content on a specific topic.

X Search Highlights

Search and extract high-value posts from X (Twitter) with engagement-based ranking.

Quick Start

# Basic search
~/.openclaw/skills/x-search-highlights/scripts/x-search.sh "Claude Code"

# With filters
~/.openclaw/skills/x-search-highlights/scripts/x-search.sh "AI Agent" 10 5 1000 markdown

Description

Extract valuable posts from X search results based on engagement metrics (likes, retweets, replies, views). Ideal for discovering trending discussions, finding expert insights, or curating content for research.

When to Use

Activate when user asks to:

  • Search X/Twitter for specific topics
  • Find "best" or "high-value" posts on a subject
  • Extract posts with engagement data
  • Curate content from X discussions
  • Discover trending discussions

Trigger phrases:

  • "Search X for [topic]"
  • "Find best posts about [topic] on Twitter"
  • "Get highlights from X search"
  • "Extract valuable tweets"

Input Parameters

Parameter Type Default Description
topic string required Search query (e.g., "Claude Code", "AI Agent")
maxResults number 5 Maximum number of posts to return
minLikes number 0 Minimum likes threshold (filter low-engagement)
scrollTimes number 3 Number of scroll iterations (more = more candidates)
sortBy string "engagement" Sort method: engagement, likes, views, recent
outputFormat string "markdown" Output format: markdown, json, summary

Output Format

Markdown (default)

## 1. [Post Title/Summary]

- **标题**：[Content summary]
- **日期**：YYYY-MM-DD
- **标签**：#tag1 #tag2
- **亮点**：🎯 [Key insight] 💡 [Unique perspective]
- **互动**：X 回复 · X 转发 · X 点赞 · X 浏览
- **链接**：[点击阅读原文](URL)

JSON

{
  "total": 10,
  "posts": [
    {
      "text": "...",
      "author": "...",
      "likes": 1000,
      "retweets": 200,
      "views": 50000
    }
  ]
}

Workflow

  1. Open search page: Navigate to X search with query
  2. Load content: Scroll N times to collect candidate posts
  3. Extract data: Parse DOM for post content and engagement metrics
  4. Rank and filter: Calculate engagement scores, apply filters
  5. Format output: Return results in requested format

Core Algorithm

Engagement Score:

Score = (likes × 2) + (retweets × 5) + (views × 0.01)

Weight Rationale:

  • Retweets (×5): Strongest signal (public sharing)
  • Likes (×2): Approval signal (low barrier)
  • Views (×0.01): Reach indicator (easily inflated)

Dependencies

  • bb-browser ≥ 0.11.2
  • Chrome/Chromium browser
  • X.com login state (in bb-browser profile)

Installation

# Via ClawHub (after publishing)
clawhub install x-search-highlights

# Or clone from GitHub
git clone https://github.com/Ghostwritten/x-search-highlights.git ~/.openclaw/skills/x-search-highlights

Usage Examples

# Search for "OpenClaw"
scripts/x-search.sh "OpenClaw"

# Get 10 posts with min 1000 likes
scripts/x-search.sh "AI Agent" 10 3 1000

# JSON output
scripts/x-search.sh "RAG" 20 5 0 json

Troubleshooting

No posts found:

  • Check bb-browser is running: bb-browser status
  • Verify X.com login state
  • Try different search keywords

JSON parsing errors:

  • Ensure bb-browser version ≥ 0.11.2
  • Check Chrome/Chromium is accessible

Rate limits:

  • Reduce scrollTimes parameter
  • Add delays between operations

Limitations

  • X lazy loading limits initial results
  • Bookmark data not available via scraping
  • Rate limits may affect large-scale scraping
  • Search quality depends on X algorithm