feat: add x_read_article tool for reading full X Article content

[nj-io]

Summary

New scrapeArticle() scraper and x_read_article MCP tool for reading the full content of X Articles (long-form posts).

What it does

• Accepts a tweet URL or direct article URL
• For tweet URLs, discovers the /article/ link from the page
• For quote tweets where the article belongs to the quoted author (not the tweet author), falls back to clicking article-cover-image and following the navigation
• Scrolls through the article to load lazy content
• Returns cleaned text with header/footer noise stripped

Return shape

{
  "title": "From Hierarchy to Intelligence",
  "author": "jack",
  "handle": "jack",
  "text": "At Sequoia, we see that speed is the best predictor...",
  "images": ["https://pbs.twimg.com/media/..."],
  "url": "https://x.com/jack/article/2039003879841362278"
}

Architecture

Follows the standard scraper pattern:

• scrapeArticle() in src/scrapers/twitter/index.js
• Re-exported through src/scrapers/index.js
• Wrapped as x_read_article() in src/mcp/local-tools.js
• Tool definition in src/mcp/server.js (Articles section)

Pairs with

This tool pairs well with x_get_likes (PR #13) — when liked tweets contain articles, the article metadata includes a tweetUrl that can be passed directly to x_read_article to fetch the full content.

Test plan

• x_read_article with direct article URL returns full content
• x_read_article with tweet URL discovers and reads the article
• x_read_article with quote-tweet URL (article from quoted author) clicks through correctly
• Article text is clean — no author metadata, timestamps, or engagement numbers in output
• Profile images filtered from returned images array

🤖 Generated with Claude Code

[vercel]

@nj-io is attempting to deploy a commit to the kaivocmenirehtacgmailcom's projects Team on Vercel.

A member of the Team first needs to authorize it.

[nj-io]

Superseded — resubmitting as clean PRs from current codebase.