Search X Skill for OpenClaw

Search X/Twitter in real-time using xAI's Responses API with the x_search tool. Get actual tweets with citations, engagement context, and direct links. No scraping, no cookies - just the xAI API.

What it does

• Responses API with x_search - server-side search orchestration with automatic citations
• Four search modes - keyword, semantic, user search, and thread fetch
• Multi-agent research - Grok 4.20 multi-agent model for complex cross-topic analysis
• Real-time tweet search - find tweets, trends, and discussions as they happen
• Thread reader - follow reply chains and expand full threads
• Trend analysis - see what's trending around any topic
• Profile lookup - see what any account has been posting about
• Image search - Grok-4.20 returns image results from X (screenshots, memes, photos)
• Prompt caching - 90% input cost reduction on repeated searches
• Batch API - bulk search processing for monitoring and research workflows
• Time and handle filtering - search any time range, specific accounts, or exclude noise
• Multiple output formats - full results, compact, links-only, or JSON

Quick start

Install the skill

git clone https://github.com/mvanhorn/clawdbot-skill-search-x.git ~/.openclaw/skills/search-x

Set up your API key

Get a key from console.x.ai, then:

export XAI_API_KEY="xai-YOUR-KEY"

Example chat usage

• "Search X for what people are saying about Claude Code"
• "Find tweets from @remotion_dev in the last week"
• "What's trending on Twitter about AI today?"
• "Read this thread from @karpathy about transformers"
• "Who is @levelsio on X? What do they post about?"
• "Show me images people are sharing about the new MacBook on X"
• "Give me a comprehensive analysis of the AI safety debate on X this month"

CLI usage

# Basic search
node scripts/search.js "AI video editing"

# Time-scoped
node scripts/search.js --days 7 "breaking news"
node scripts/search.js --days 1 "trending today"

# Handle filters
node scripts/search.js --handles @elonmusk,@OpenAI "AI announcements"
node scripts/search.js --exclude @bots "real discussions"

# Output formats
node scripts/search.js --compact "topic"
node scripts/search.js --links-only "topic"
node scripts/search.js --json "topic"

# Image search (requires grok-4.20 model)
node scripts/search.js --model grok-4.20-beta-0309-reasoning "memes about programming"

# Multi-agent research
node scripts/search.js --model grok-4.20-multi-agent-beta-0309 "comprehensive AI safety analysis"

Models

| Model | Context | Price (in/out) | Best For | |-------|---------|----------------|----------| | grok-4.20-multi-agent-beta-0309 | 2M | $2 / $6 | Complex multi-step research | | grok-4.20-beta-0309-reasoning | 2M | $2 / $6 | Deep analysis with reasoning, images | | grok-4-1-fast-reasoning | 2M | $0.20 / $0.50 | Fast search with reasoning | | grok-4-1-fast-non-reasoning (default) | 2M | $0.20 / $0.50 | Quick searches (cheapest) |

How it works

Uses xAI's Responses API (POST /v1/responses) with the x_search tool. This is a first-party xAI tool with access to the full X firehose - not a scraper. The Responses API handles all search orchestration server-side: Grok decides when to search, executes the search, and returns structured results with URL citations. You can combine x_search with web_search and code_interpreter in the same request. See SKILL.md for full documentation of all six search modes, prompt caching, and the Batch API.

Environment variables

| Variable | Required | Default | Description | |----------|----------|---------|-------------| | XAI_API_KEY | Yes | - | Your xAI API key | | SEARCH_X_MODEL | No | grok-4-1-fast-non-reasoning | Default model | | SEARCH_X_DAYS | No | 30 | Default time window |

License

MIT