Supercharge Your Agent's Web Search with Tavily

Stop getting garbage search results. If you've ever watched your AI agent struggle with generic web search — returning SEO-stuffed articles, paywalled content, or irrelevant snippets — Tavily fixes that.

The Tavily Web Search skill brings AI-optimized search to Clawdbot. Unlike traditional search engines, Tavily is built specifically for AI agents: it returns clean, relevant content without the cruft.

Who Needs This?

Researchers who want their agent to find accurate, current information
News junkies who need agents that track current events
Anyone tired of agents hallucinating when they should be searching

Installation

npx clawdhub@latest install tavily-search

Then get your API key from tavily.com (free tier available) and add it to your environment:

# Add to your shell profile (~/.zshrc, ~/.bashrc)
export TAVILY_API_KEY="tvly-your-key-here"

# Or add to your Clawdbot .env file
echo "TAVILY_API_KEY=tvly-your-key-here" >> ~/.openclaw/workspace/.env

Restart your terminal or Clawdbot gateway after setting the key.

Usage Examples

Basic Search

node ~/.openclaw/workspace/skills/tavily-search/scripts/extract.mjs "https://example.com/article"

Get More Results

node ~/.openclaw/workspace/skills/tavily-search/scripts/extract.mjs "https://example.com/article"

Deep Research Mode

For complex questions needing comprehensive answers:

node ~/.openclaw/workspace/skills/tavily-search/scripts/extract.mjs "https://example.com/article"

News Search

Get current events with optional time filtering:

node ~/.openclaw/workspace/skills/tavily-search/scripts/extract.mjs "https://example.com/article"

Extract Content from URL

Pull clean content from any webpage:

node ~/.openclaw/workspace/skills/tavily-search/scripts/extract.mjs "https://example.com/article"

Pro Tips

Use --deep for research questions — it's slower but returns more comprehensive results. Great for "compare X vs Y" or "how does X work" queries.
Combine with news topic for current events — --topic news --days 3 ensures you're getting fresh information, not stale articles.
The extract feature is underrated — when you have a specific URL but want clean markdown instead of HTML soup, use extract.mjs.
Free tier is generous — Tavily offers 1,000 searches/month free. Enough for personal use.

Troubleshooting

"API key not found": Make sure TAVILY_API_KEY is exported in your shell AND Clawdbot's environment. Restart the gateway after adding it.

Security warnings: You might see warnings about "env-harvesting" — this is expected behavior. The skill reads your API key to authenticate with Tavily. Review the source code if concerned; it's a simple 2-file skill.

Why Tavily Over Built-in Search?

Clawdbot's built-in web_search tool uses Brave Search, which works fine. But Tavily is specifically optimized for AI consumption:

Pre-extracts relevant content (no scraping needed)
Filters out SEO spam and low-quality sources
Returns structured data agents can parse easily
Deep search mode for research-grade queries

For quick lookups, use built-in search. For research, use Tavily.

Skill by @arun-8687 • 53 ⭐ • 21k+ downloads