---
name: x-ai-articles
description: Scrape high-quality AI long-form articles and threads from X (Twitter) using OpenClaw browser. Generates a daily digest of the most popular AI-related deep dives, threads (🧵), articles, breakdowns, and tutorials sorted by engagement. Triggers: /aix command, "X AI articles", "Twitter AI threads", "AI trending posts", "top AI tweets today", "AI长文", "X上的AI文章", or requests for daily AI content digest from X/Twitter. Requires OpenClaw browser with an active X login session (profile=openclaw).
---

# X AI Articles Scraper

Scrape high-quality, long-form AI articles and threads from X (Twitter) and deliver a curated daily digest.

## Prerequisites

- OpenClaw browser with active X/Twitter login (profile=openclaw)
- Account: @Clovericbot (joannagrant14@gmail.com) or any logged-in X account

## Workflow

### Step 1: Open X Advanced Search

Use the OpenClaw browser (profile=openclaw) to search X with this query template:

```
(AI OR LLM OR AGI OR Claude OR GPT OR OpenAI OR Anthropic OR "coding agent" OR "AI agent")
(thread OR 🧵 OR article OR "deep dive" OR "here's what" OR "lessons" OR "breakdown" OR "tutorial" OR "guide" OR "wrote about" OR "published")
min_faves:200
since:{YYYY-MM-DD_yesterday}
until:{YYYY-MM-DD_today}
```

URL pattern:
```
https://x.com/search?q={URL_ENCODED_QUERY}&src=typed_query&f=top
```

Sort by "Top" tab to get engagement-ranked results.

### Step 2: Extract Posts via Snapshot

Take a browser snapshot (not screenshot) of the search results page. Extract from each article element:

- **Author**: Display name + @handle
- **Content**: Full tweet text (click "Show more" if truncated)
- **Engagement**: Likes, reposts, bookmarks, views
- **URL**: Construct from `https://x.com/{handle}/status/{status_id}`
- **Type**: Thread (🧵), Article, Quote-tweet with article, etc.

### Step 3: Filter for Quality

Keep only posts that are **substantive long-form content**:
- ✅ Thread/🧵 with educational content
- ✅ Article links with analysis
- ✅ Deep dives, tutorials, breakdowns, lessons learned
- ✅ Product launches with detailed explanations
- ❌ One-liner hot takes (even if high engagement)
- ❌ Memes, jokes, engagement bait
- ❌ Ads / promoted content
- ❌ Pure news headlines without analysis

### Step 4: Format Output

Output as a ranked list sorted by engagement (views × quality):

```
## 📖 X 热门 AI 长文 | {date}

**1️⃣ {Title/Summary}**
@{handle} | ❤️ {likes} | 🔖 {bookmarks} | 👁️ {views}
{2-3 sentence summary of the content}
https://x.com/{handle}/status/{id}

**2️⃣ ...
```

Target: 5-10 high-quality posts per daily digest.

### Step 5: Deliver

Send the formatted digest to the user via their preferred channel (Telegram/Discord/etc).

## Adjustable Parameters

- **Keywords**: Add domain-specific terms (e.g., "RAG", "fine-tuning", "computer use")
- **min_faves**: Lower (100) for niche topics, higher (500) for mainstream
- **Time range**: Default is past 24 hours; can be adjusted for weekly digests
- **Language**: Add `lang:en` or `lang:ja` filter for language-specific results

## Cron Setup (Optional)

To run as a daily automated task, create a cron job with:
- Schedule: Daily at a preferred time (e.g., 18:00 local)
- Payload: agentTurn with the scraping workflow
- Delivery: announce to user's chat

## Troubleshooting

- **No results**: Lower min_faves threshold or broaden keywords
- **Login required**: Re-authenticate via OpenClaw browser (profile=openclaw) → x.com
- **Rate limited**: Wait 5 minutes between searches; X may throttle automated searches