twitter-api-v2（twitter API V2 搜索）

X (Twitter) API

Interact with X via API v2. Pure Python stdlib — no dependencies.

Features

Public data (bearer token):

• User profiles (bio, followers, tweets count)
• User timelines (with filters: exclude retweets/replies)
• Conversation threads (self-threads by same author)
• Search recent tweets (last 7 days)
• Single/batch tweet lookup
• Tweet metrics (likes, retweets, replies)

User context (OAuth 2.0):

• Bookmarks
• Liked tweets
• Post tweets

Setup

See SETUP.md for credentials setup (bearer token + optional OAuth 2.0).

Usage

User Info

# Get user profile (bio, followers, verification status)
python3 {baseDir}/scripts/x.py user steipete
python3 {baseDir}/scripts/x.py user elonmusk

Timeline

# Get user's recent tweets
python3 {baseDir}/scripts/x.py timeline steipete --max 50

# Exclude retweets
python3 {baseDir}/scripts/x.py timeline steipete --max 100 --exclude retweets

# Exclude both retweets and replies
python3 {baseDir}/scripts/x.py timeline steipete --max 100 --exclude retweets,replies

Threads

# Get conversation thread (self-thread by same author)
python3 {baseDir}/scripts/x.py thread https://x.com/steipete/status/123456
python3 {baseDir}/scripts/x.py thread 123456789

# Note: Only searches last 7 days (recent search). For older threads, need full-archive search.

Search

# Search recent tweets (last 7 days)
python3 {baseDir}/scripts/x.py search "OpenClaw" --max 20
python3 {baseDir}/scripts/x.py search "from:steipete OpenAI" --max 50

# Advanced search operators work (from:, to:, -filter:retweets, etc.)

Tweet Lookup

# Single tweet
python3 {baseDir}/scripts/x.py tweet 2025504475100614695
python3 {baseDir}/scripts/x.py tweet https://x.com/steipete/status/2025504475100614695

# Multiple tweets (batch)
python3 {baseDir}/scripts/x.py tweets 123456 789012 345678

Bookmarks (OAuth Required)

# One-time auth setup
python3 {baseDir}/scripts/x.py auth

# Get your bookmarks
python3 {baseDir}/scripts/x.py bookmarks --max 100

Likes (OAuth Required)

# Get user's liked tweets (requires OAuth)
python3 {baseDir}/scripts/x.py likes steipete --max 50

Posting (OAuth Required)

# Post a tweet
python3 {baseDir}/scripts/x.py post "Hello from OpenClaw! 🦞"

# Note: 280 character limit enforced by API

Output Format

All commands output clean, readable text:

👤 Peter Steinberger 🦞 (@steipete)
   ID: 25401953
   Bio: Polyagentmorous ClawFather...
   👥 382,854 followers · 2,241 following · 135,096 tweets
   https://x.com/steipete

🐦 @VictorKnox99 I think my stance on using AI here is pretty clear.
   📅 2026-02-22 09:34:33
   ❤️  5  🔁 0  💬 0
   https://x.com/steipete/status/2025504475100614695

Common Use Cases

Research user activity:

# Get someone's recent thoughts on a topic
python3 {baseDir}/scripts/x.py timeline steipete --max 100 --exclude retweets | grep -i openai

Monitor discussions:

# Search for OpenClaw mentions
python3 {baseDir}/scripts/x.py search "OpenClaw" --max 100

Extract threads:

# Get full thread for analysis
python3 {baseDir}/scripts/x.py thread <tweet_url> > thread.txt

Archive bookmarks:

# Export bookmarks to file
python3 {baseDir}/scripts/x.py bookmarks --max 100 > bookmarks.txt

Rate Limits

X API enforces rate limits per 15-minute window:

Headers in responses:

• x-rate-limit-limit — max requests allowed
• x-rate-limit-remaining — requests left in window
• x-rate-limit-reset — Unix timestamp when limit resets

Design tips:

• Cache results locally (deduplication helps but isn't guaranteed)
• Add delays between requests (avoid hitting limits)
• Use batch endpoints when possible (tweets for multiple IDs)

Billing & Costs

Pay-per-usage model:

• Buy credits in Developer Console
• Charged per unique tweet/user returned
• 24-hour deduplication — same tweet fetched twice in a UTC day = charged once
• Set spending limits to control costs
• Monitor usage: /2/usage/tweets endpoint or console

What's billable:

• Tweet lookups
• Search results
• Timeline requests
• Bookmarks retrieval

What's NOT billable:

• Failed requests (4xx/5xx errors)
• Re-fetching same tweet within 24h UTC window (deduplicated)

See references/pricing.md for detailed pricing info.

Search Operators

X API search supports advanced operators:

Example:

# OpenClaw tweets from verified users, excluding retweets
python3 {baseDir}/scripts/x.py search "OpenClaw filter:verified -filter:retweets" --max 50

Limitations

Recent search only:

• search and thread commands only cover last 7 days
• Full-archive search requires Pro tier ($5,000/mo) or pay-per-usage

Bookmarks/likes require OAuth:

• Bearer token alone can't access private user data
• Need OAuth 2.0 user context (one-time setup)

No Watch Later equivalent:

• X API doesn't expose "Watch Later" or similar lists
• Use bookmarks as alternative

Rate limits still apply:

• Pay-per-usage removes monthly caps but rate limits per 15-min window remain
• Design for retries with exponential backoff

Troubleshooting

403 Forbidden:

• Check credentials file exists and has valid bearer_token
• Ensure endpoint doesn't require OAuth (bookmarks, likes, posting need OAuth)

429 Too Many Requests:

• Hit rate limit — wait until x-rate-limit-reset time
• Reduce --max values
• Add delays between requests

No results found:

• Recent search only covers 7 days — thread/search may miss older content
• User may have no tweets or privacy settings blocking access

Authentication errors (OAuth):

• Ensure oauth2.json has correct client_id/client_secret
• Re-run x.py auth to refresh tokens
• Check scopes in Developer Portal match required scopes

Notes

• Pure stdlib — no pip dependencies, uses urllib + json
• Works offline — caches nothing, always fetches fresh data
• Thread-safe — stateless CLI, safe for parallel invocations
• Deduplication — X API deduplicates within 24h UTC day (soft guarantee)
• Output encoding — UTF-8, handles emoji and international characters