Recherche Tavily

Offre des capacités de recherche web optimisées par l'IA via l'API Tavily, incluant l'extraction de contenu et la génération de rapports de recherche.

Spar Skills Guide Bot
Data & IAIntermédiaire
1022/07/2026
Claude CodeCursorWindsurf
#web-search#tavily#research#content-extraction

Recommandé pour


name: tavily-search description: Provides web search capabilities using Tavily AI-optimized search API. Supports searching the web for relevant results, extracting content from URLs, crawling websites, and generating AI-powered research reports.

Tavily Search

This skill provides powerful web search capabilities using the Tavily API, which is specifically optimized for AI applications.

API Configuration

API Key: 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

Available Commands

/search - Web Search

Search the web for relevant information using Tavily's LLM-optimized search API.

Usage:

/search current React best practices
/search quantum computing applications
/search latest AI agent developments

What it does:

  • Returns relevant search results with content snippets
  • Provides metadata about sources
  • Filters for relevance and quality

/research - Deep Research

Generate comprehensive AI-powered research reports on any topic.

Usage:

/research AI agent architecture and trends
/research comparison of frontend frameworks 2024
/research best practices for Python API design

What it does:

  • Compiles information from multiple sources
  • Provides structured JSON output
  • Includes citations and references
  • Supports saving reports to files

/extract - Content Extraction

Extract clean content from specific URLs.

Usage:

/extract https://example.com/blog/post
/extract https://docs.example.com/guide

What it does:

  • Returns webpage content in markdown/text format
  • Preserves structure and formatting
  • Removes ads and clutter

/crawl - Website Crawling

Crawl websites and save as local markdown files.

Usage:

/crawl https://docs.example.com
/crawl https://example.com/api --save ./docs

What it does:

  • Downloads entire websites as markdown
  • Preserves directory structure
  • Ideal for building knowledge bases

When to Use This Skill

Use this skill when the user:

  • Asks to search the web for current information
  • Wants to research a topic comprehensively
  • Needs to extract content from specific URLs
  • Wants to crawl documentation or knowledge bases
  • Asks about recent developments or trends
  • Needs citations for sources

Examples

Example 1: Search for Current Best Practices

User: "What are the current React 19 best practices?"
→ /search React 19 new features best practices

Example 2: Research Topic with Sources

User: "I need a comprehensive report on vector databases"
→ /research vector database comparisons and save to vector-db-report.json

Example 3: Extract Documentation

User: "Can you get the content from this documentation page?"
→ /extract https://docs.example.com/guide/overview

Example 4: Build Knowledge Base

User: "Download the React documentation as markdown"
→ /rawl https://react.dev --save ./react-docs

Best Practices

  1. Be Specific: Use detailed search queries for better results

    • Good: "React 19 server actions best practices"
    • Bad: "React stuff"
  2. Use Research for Complex Topics: For comprehensive understanding

    • /research compiles multiple sources
    • Includes citations for verification
  3. Extract for Specific Content: When you need the full content of a page

    • Better than copy-paste for formatted content
    • Preserves markdown structure
  4. Crawl for Documentation: Build local knowledge bases

    • Great for offline documentation access
    • Preserves site structure

API Configuration in Settings

To configure Tavily API key in Claude Code settings:

Windows:

notepad %USERPROFILE%\.claude\settings.json

macOS/Linux:

nano ~/.claude/settings.json

Add the following configuration:

{
  "env": {
    "TAVILY_API_KEY": "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"
  }
}

Troubleshooting

Q: Search returns no results

  • Check API key is correctly configured
  • Verify network connectivity
  • Try rephrasing the search query

Q: Rate limiting errors

  • Tavily has rate limits based on your plan
  • Wait a moment before retrying
  • Consider upgrading your Tavily plan

Q: Content extraction fails

  • Ensure URL is accessible
  • Check if the site blocks automated requests
  • Try crawling instead of extracting

Related Resources

Skills similaires