Tweet Series Extractor

Tweet Series Extractor

Extract recurring tweet series from X/Twitter with content, links, and engagement metrics.

Prerequisites

Load Chrome tools before use:

ToolSearch: select:mcp__claude-in-chrome__tabs_context_mcp
ToolSearch: select:mcp__claude-in-chrome__navigate
ToolSearch: select:mcp__claude-in-chrome__get_page_text
ToolSearch: select:mcp__claude-in-chrome__read_page
ToolSearch: select:mcp__claude-in-chrome__javascript_tool

Workflow

1. Initialize Browser

mcp__claude-in-chrome__tabs_context_mcp(createIfEmpty: true)

2. Navigate to Seed Tweet

mcp__claude-in-chrome__navigate(url: "https://x.com/username/status/ID", tabId: <id>)

3. Extract Tweet Content

mcp__claude-in-chrome__get_page_text(tabId: <id>)

Returns: Title, URL, full article text, engagement metrics inline.

4. Extract Links from Tweet

// Use javascript_tool with this code:
var links = [];
document.querySelectorAll('article a').forEach(function(a) {
  if (a.href && a.href.length > 10) links.push(a.href);
});
links.slice(4, 50).join(' | ')

• slice(4, 50) skips profile/header links
• Returns pipe-separated URLs

5. Find Similar Tweets on Profile

Navigate to profile and scroll:

mcp__claude-in-chrome__navigate(url: "https://x.com/username", tabId: <id>)
mcp__claude-in-chrome__javascript_tool(text: "window.scrollBy(0, 2000); 'scrolled'")
mcp__claude-in-chrome__read_page(tabId: <id>, depth: 12)

Look for articles with matching title patterns (e.g., "Welcome to [Date]"). Extract tweet URLs from href="/username/status/ID" links.

6. Collect Each Tweet

For each found tweet URL, repeat steps 2-4.

Parsing Engagement Metrics

Metrics appear in page text after the timestamp:

6:12 AM · Jan 24, 2026·33.7K Views
80 68 594 163

Format: replies reposts likes bookmarks

Also found in group elements:

group "80 replies, 68 reposts, 594 likes, 163 bookmarks, 33738 views"

Output JSON Structure

{
  "author": {
    "name": "Display Name",
    "handle": "@username",
    "profile_url": "https://x.com/username"
  },
  "series": "Pattern description",
  "collected_at": "YYYY-MM-DD",
  "tweets": [
    {
      "date": "Month DD, YYYY",
      "url": "https://x.com/username/status/ID",
      "engagement": {
        "replies": 0,
        "reposts": 0,
        "likes": 0,
        "bookmarks": 0,
        "views": 0
      },
      "content": "Full tweet text...",
      "links": [
        {"text": "anchor text", "url": "https://..."}
      ]
    }
  ]
}

Tips

• Scroll profile multiple times to load more tweets
• Tweet URLs follow pattern: /username/status/ID
• Articles are X's long-form format with embedded links
• Use read_page with depth: 12 for full tweet visibility
• Parse link text from the generic elements near link refs