Fetch User Timeline

# Fetch latest tweets from a user (supports pagination, MAX_PAGES=20)
python3 scripts/fetch_tweet.py --user <username> --limit 300

Fetch Chinese Platform Content

# Auto-detects platform from URL
python3 scripts/fetch_china.py --url "https://weibo.com/..."     # Weibo
python3 scripts/fetch_china.py --url "https://bilibili.com/..."  # Bilibili
python3 scripts/fetch_china.py --url "https://csdn.net/..."      # CSDN
python3 scripts/fetch_china.py --url "https://mp.weixin.qq.com/..." # WeChat (no Camofox needed!)

| Platform | Status | Notes | |----------|--------|-------| | WeChat Articles | ✅ | Uses web_fetch directly, no Camofox | | Weibo | ✅ | Camofox renders JS | | Bilibili | ✅ | Video info + stats | | CSDN | ✅ | Articles + code blocks | | Zhihu / Xiaohongshu | ⚠️ | Needs cookie import for login |

Google Search (Zero API Key)

# Python
from scripts.camofox_client import camofox_search
results = camofox_search("your search query")
# Returns: [{"title": "...", "url": "...", "snippet": "..."}, ...]

# CLI
python3 scripts/camofox_client.py "your search query"

Uses Camofox browser to search Google directly. No Brave API key needed, no cost.

Camofox Setup

What is Camofox?

Camofox is an anti-detection browser service based on Camoufox (a Firefox fork with C++ level fingerprint masking). It bypasses:

• Cloudflare bot detection
• Browser fingerprinting
• JavaScript challenges

Installation

Option 1: OpenClaw Plugin

openclaw plugins install @askjo/camofox-browser

Option 2: Manual Install

git clone https://github.com/jo-inc/camofox-browser
cd camofox-browser
npm install && npm start

Verify

curl http://localhost:9377/health
# Should return: {"status":"ok"}

REST API

# Create tab
POST http://localhost:9377/tabs
Body: {"userId":"test", "sessionKey":"test", "url":"https://example.com"}

# Get page snapshot
GET http://localhost:9377/tabs/<TAB_ID>/snapshot?userId=test

# Close tab
DELETE http://localhost:9377/tabs/<TAB_ID>?userId=test

From Agent Code

from scripts.fetch_tweet import fetch_tweet

result = fetch_tweet("https://x.com/user/status/123456")
tweet = result["tweet"]

# Regular tweet
print(tweet["text"])
print(f"Likes: {tweet['likes']}, Views: {tweet['views']}")

# X Article (long-form)
if tweet.get("is_article"):
    print(tweet["article"]["title"])
    print(tweet["article"]["full_text"])

# Links found in replies
for reply in result.get("replies", []):
    for link in reply.get("links", []):
        print(link)

Output Format

{
  "url": "https://x.com/user/status/123",
  "username": "user",
  "tweet_id": "123",
  "tweet": {
    "text": "Tweet content...",
    "author": "Display Name",
    "screen_name": "username",
    "likes": 100,
    "retweets": 50,
    "bookmarks": 25,
    "views": 10000,
    "replies_count": 30,
    "created_at": "Mon Jan 01 12:00:00 +0000 2026",
    "is_note_tweet": false,
    "is_article": true,
    "article": {
      "title": "Article Title",
      "full_text": "Complete article content...",
      "word_count": 4847
    }
  },
  "replies": [
    {
      "author": "@someone",
      "text": "Reply text...",
      "likes": 5,
      "links": ["https://github.com/..."],
      "thread_replies": [{"text": "Nested reply..."}]
    }
  ]
}

File Structure

x-tweet-fetcher/
├── SKILL.md                    # This file
├── README.md                   # GitHub page with full docs
├── scripts/
│   ├── fetch_tweet.py          # Main fetcher (tweet + replies + timeline)
│   ├── fetch_china.py          # Chinese platform fetcher
│   ├── camofox_client.py       # Camofox REST API client + camofox_search()
│   └── x-profile-analyzer.py   # User profile analysis (AI-powered)
└── CHANGELOG.md

Requirements

• Basic: Python 3.7+, no external packages, no API keys
• Advanced: Camofox running on localhost:9377
• Profile Analyzer: MiniMax M2.5 API key (for AI analysis)

X-Tracker: Tweet Growth Monitoring

Track your tweets' growth and detect viral moments — inspired by semiconductor ETCH Endpoint Detection.

Quick Start

# Add a tweet to track
python3 scripts/tweet_growth_cli.py --add "https://x.com/user/status/123" "my launch tweet"

# List all tracked tweets
python3 scripts/tweet_growth_cli.py --list

# Run sampling (new tweets <48h, every 15min)
python3 scripts/tweet_growth_cli.py --run --fast

# Run sampling (all tweets, hourly)
python3 scripts/tweet_growth_cli.py --run --normal

# Generate analysis report
python3 scripts/tweet_growth_cli.py --report 123456789

# Report with topic cross-analysis
python3 scripts/tweet_growth_cli.py --report 123456789 --cross

Cron Integration

# Dual-frequency sampling
*/15 * * * * python3 tweet_growth_cli.py --run --fast    # New tweets (<48h)
0 * * * *    python3 tweet_growth_cli.py --run --normal  # All tweets (hourly)

Detection Algorithm

| Component | Method | Purpose | |-----------|--------|---------| | Derivative detection | dV/dt per hour | Spot sudden acceleration | | Sliding window | 5-sample moving average | Filter noise | | Multi-signal fusion | views×1 + likes×1 + bookmarks×1.5 + RT×3 | Weighted composite score | | Burst confirmation | 3 consecutive windows above threshold | Prevent false positives | | Surge override | Single window +100%/h | Catch massive spikes | | Saturation | 6 samples < 2%/h growth | Detect long tail | | Propagation | RT-per-1k-views ratio | Influencer vs algorithm driven |

Output Example

🎯 Burst detected at 2026-03-14 08:45
   - Growth rate: 156%/h
   - Composite score: +2847 (views +1200, RT +89, likes +234)
   - Propagation: 12.3 RT/1k views (influencer-driven)

How It Works

• Basic tweets: FxTwitter public API
• Replies/timelines: Camofox → Nitter (privacy-respecting X frontend)
• Chinese platforms: Camofox renders JS → extracts content
• Google search: Camofox opens Google → parses results
• X-Tracker: FxTwitter API + derivative detection + burst confirmation

| Backend | Deps | Speed | Features | |---------|------|-------|----------| | nitter | None (stdlib only) | ⚡ Fast | Timeline, search, replies, profile, mentions | | browser | Playwright/Chromium | 🐢 Slower | Everything above + Lists + Articles + fetch_china | | auto (default) | Best available | ⚡→🐢 | Tries nitter first, falls back to browser |

OpenClaw users: Playwright + Chromium are built-in. --backend auto just works — no extra install needed.

📊 Capabilities

| Feature | Backend | Output | |---------|---------|--------| | Single tweet | FxTwitter (always) | text, stats, media, quotes | | Reply comments | nitter / browser | threaded comment list | | User timeline | nitter / browser | paginated tweet list | | @mentions monitor | nitter / browser | incremental new mentions | | Keyword search | nitter / browser | real-time tweet stream | | X Lists | browser only | list member tweets | | X Articles | browser only | full long-form content | | User profile analysis | nitter + LLM | MBTI, Big Five, topic graph | | WeChat article search | Sogou (direct HTTP) | title, url, author, date | | WeChat/Weibo/Bilibili | browser only | via fetch_china.py | | Tweet growth tracker | FxTwitter API | growth curves, burst detection |

For AI Agents: All output is structured JSON. Import as Python modules for direct integration. Exit codes are cron-friendly (0=nothing new, 1=new content).

🧭 Agent Waystation

x-tweet-fetcher is one of the field tools maintained from Agent Waystation — a stopover for AI agents, builders, tools, and field reports.

If you are building an agent that reads X/Twitter, bring it to the Waystation:

• What does your agent observe?
• What does it remember?
• Where does it fail?
• What field tool does it still need?

中文：如果你也在养一个会看 X/Twitter 的 Agent，欢迎带它来 Agent 驿站 报到。说说它会看什么、会记什么、最容易在哪翻车。

→ Visit Agent Waystation / Agent 驿站

🚀 Quick Start

Single tweet (zero setup)

# Works immediately — no Nitter, no browser needed
python3 scripts/fetch_tweet.py --url https://x.com/elonmusk/status/123456789

Timeline, search, replies

# Set your Nitter instance URL (for nitter/auto mode)
export NITTER_URL=http://127.0.0.1:8788

# User timeline
python3 scripts/fetch_tweet.py --user elonmusk --limit 20

# Keyword search — real-time tweets
python3 scripts/nitter_client.py --search "AI agent"

# Tweet replies
python3 scripts/fetch_tweet.py --url https://x.com/elonmusk/status/123456789 --replies

# @mentions monitoring (cron-friendly)
python3 scripts/fetch_tweet.py --monitor @yourusername

# User profile analysis
python3 scripts/x-profile-analyzer.py --user elonmusk --count 100

Lists & Articles (browser backend)

# X List — requires Playwright
python3 scripts/fetch_tweet.py --list 1455045069516357634 --backend browser

# X Article
python3 scripts/fetch_tweet.py --article https://x.com/user/article/123 --backend browser

# WeChat / Weibo / Bilibili
python3 scripts/fetch_china.py --url "https://mp.weixin.qq.com/s/..."

WeChat search (always zero-dep)

python3 scripts/sogou_wechat.py --keyword "AI Agent" --limit 5 --json

🖥️ Works with Claude Code / CC

Since x-tweet-fetcher has zero mandatory dependencies, it works perfectly in constrained environments:

| Environment | nitter mode | browser mode | Notes | |-------------|:----------:|:------------:|-------| | Claude Code (CC) | ✅ | ❌ | No browser runtime | | OpenClaw | ✅ | ✅ | Playwright built-in | | VPS (headless Linux) | ✅ | ✅* | needs pip install playwright | | Mac / Windows | ✅ | ✅ | *needs pip install playwright | | CI/CD pipelines | ✅ | ⚠️ | Possible but heavy | | Docker containers | ✅ | ⚠️ | Needs Chromium in image | | Termux (Android) | ✅ | ❌ | No Chromium |

# In Claude Code (nitter mode, zero deps):
export NITTER_URL=http://your-vps:8788
python3 scripts/fetch_tweet.py --user YuLin807 --limit 10

# In OpenClaw (auto mode, full features):
python3 scripts/fetch_tweet.py --user YuLin807 --limit 10
# → auto-detects Nitter, falls back to Playwright if needed

🔧 Self-hosted Nitter Setup

⚠️ Public Nitter instances are dead or unreliable (as of March 2026). Self-hosting is the only reliable option.

Why you need this

Twitter removed guest API access in 2023. Public Nitter instances get rate-limited because thousands of users share a few accounts. Your own instance = your own rate limits.

5-minute setup guide

1. Install dependencies

# Ubuntu/Debian
sudo apt install -y redis-server libpcre3-dev libsass-dev

# Install Nim
curl https://nim-lang.org/choosenim/init.sh -sSf | sh
export PATH=$HOME/.nimble/bin:$PATH

2. Build Nitter

git clone https://github.com/zedeus/nitter
cd nitter
nimble build -d:release
nimble scss
cp nitter.example.conf nitter.conf

3. Get X session cookies

Use a secondary account (not your main).

1. Log into X in browser → DevTools → Application → Cookies → x.com
2. Copy auth_token and ct0
3. Create sessions.jsonl:

{"name":"myaccount","auth_token":"YOUR_AUTH_TOKEN","ct0":"YOUR_CT0"}

4. Configure

[Server]
address = "127.0.0.1"  # Local only!
port = 8788

[Config]
hmacKey = "$(openssl rand -hex 32)"

[Tokens]
tokenFile = "sessions.jsonl"

5. Run & test

sudo systemctl start redis-server
./nitter

# Test
curl http://127.0.0.1:8788/YuLin807
export NITTER_URL=http://127.0.0.1:8788
python3 scripts/nitter_client.py --search "test"

Security

• Bind to 127.0.0.1 only — never expose to public internet
• Use a secondary X account — session token gives full access
• Session tokens last ~1 year

📐 How It Works

                    ┌─────────────┐
 --url              │  FxTwitter  │  ← Public API, no auth needed
                    │  (free)     │
                    └──────┬──────┘
                           │ JSON
              ┌────────────┴────────────┐
              │    --backend auto       │
              │  ┌───────┐  ┌────────┐  │       ┌──────────┐
 --user       │  │Nitter │→→│Browser │  │       │  Agent   │
 --replies    │  │(fast) │  │(full)  │  │──────▶│  (JSON)  │
 --monitor    │  │ 0 dep │  │Playwrt │  │       │          │
 --search     │  └───────┘  └────────┘  │       └──────────┘
 --list       └─────────────────────────┘
 --article
              ┌─────────────┐
 sogou_wechat │   Sogou     │  ← Direct HTTP, no API key
 fetch_china  │  (search)   │
              └─────────────┘

• Single tweets: FxTwitter — always works, zero auth
• Timeline / Replies / Search / Mentions: Self-hosted Nitter or Playwright browser
• Lists / Articles: Playwright browser (Nitter doesn't support these)
• WeChat / China platforms: Sogou search + fetch_china.py

📦 Requirements

Python 3.7+     (that's it for nitter mode)

| Mode | Extra requirement | |------|-----------------| | --backend nitter | Nothing (Python stdlib only) | | --backend browser | pip install playwright + playwright install chromium | | --backend auto | Uses whatever is available |

⏰ Cron Integration

Exit codes for automation: 0=nothing new, 1=new content, 2=error.

# Check mentions every 30 min
*/30 * * * * NITTER_URL=http://127.0.0.1:8788 python3 fetch_tweet.py --monitor @username

# Discover tweets daily
0 9 * * * python3 nitter_client.py --search "AI Agent" >> ~/discoveries.jsonl

🤝 Contributing

Issues and PRs welcome! Core platforms:

• X/Twitter — Nitter + Playwright backends
• WeChat articles — Sogou search

Other platforms welcome as community PRs.

🙏 Acknowledgments

• Nitter by zedeus (12.6k ⭐) — self-hosted Twitter frontend
• FxTwitter — public API for single tweet data
• Playwright — browser automation for full-feature coverage
• **[OpenClaw](https://github.c