tavily-best-practices: a skill that helps Claude write Tavily search, extract, crawl and research code

Tavily's skill gives Claude the Tavily SDK reference, so it writes working search, extract, crawl, map and research code for your app or agent.

Installs: 21.4K; on skills.sh
Stars: 491; on GitHub
Files: 6; besides SKILL.md
Updated: Oct 2026; last repo commit

The skill is documentation for the Tavily search API. It shows how to set up the Python and JavaScript clients, which method fits a job (search(), extract(), crawl(), map() or research()), and the key parameters for each. Six reference files go deeper on the SDK, search, extract, crawl, research and framework integrations.

It triggers when you ask a coding assistant to build web search, content extraction, crawling or research into an agent workflow, a RAG system or an autonomous agent. The code it writes reads your TAVILY_API_KEY from the environment.

This skill helps write code. It does not search the web itself. Tavily's other skills in the same repo, such as tavily-search, use the Tavily CLI for that.

Good for

Developers adding web search to an AI agent or RAG system
Teams building site crawlers, extractors or research pipelines on Tavily
Python and JavaScript developers using LangChain, LlamaIndex, CrewAI or the Vercel AI SDK with Tavily

Install tavily-best-practices

Claude Code

Copy this prompt.

Install the tavily-best-practices skill in Claude Code. Run this command:

npx skills add https://github.com/tavily-ai/skills --skill tavily-best-practices

Then check that it loaded and tell me how to use it.

2
Paste it into Claude Code. It shows the change and asks before running it.

Requires Node.js.

From its listing on skills.sh. Source.

What's in the skill

When it loads

Build production-ready Tavily integrations with best practices baked in. Reference documentation for developers using coding assistants (Claude Code, Cursor, etc.) to implement web search, content extraction, crawling, and research in agentic workflows, RAG systems, or autonomous agents.

Sections in SKILL.md

Installation
Client Initialization
Choosing the Right Method
Quick Reference
search() - Web Search
extract() - URL Content Extraction
crawl() - Site-Wide Extraction
map() - URL Discovery
research() - AI-Powered Research
Detailed Guides

Key facts

Author: Tavily (tavily-ai)
License: MIT
Repository: tavily-ai/skills
Installs on skills.sh: 21,423 as of 2026-11-02
GitHub stars: 491 as of 2026-11-02
Last repo commit: 2026-10-30
Files in the skill folder: 7 files including SKILL.md (6 reference guides)
skills.sh audits: Gen Agent Trust Hub pass, Socket pass, Snyk warn
Skill type: SDK reference docs; the README labels it SDK
Needs: A Tavily API key in the TAVILY_API_KEY environment variable for the code it writes

What it includes

pip install tavily-python or npm install @tavily/core
Shows TavilyClient and AsyncTavilyClient, the TAVILY_API_KEY variable and an optional project_id for usage tracking.
Runs code · Installation and client setup
Web search for custom agents
Key parameters: query (under 400 characters), max_results, search_depth (ultra-fast, fast, basic or advanced), include_domains, exclude_domains and time_range.
Network · search()
Content from specific URLs
Takes up to 20 URLs, with extract_depth, a query and chunks_per_source from 1 to 5.
Network · extract()
Site-wide extraction
Key parameters: url, max_depth, max_breadth, limit, instructions, chunks_per_source, select_paths and exclude_paths.
Network · crawl()
URL discovery on a site
Lists the URLs of a site so you can extract selected ones.
Network · map()
End-to-end research with an AI-written report
Models mini, pro or auto, with polling or streaming, an optional output schema and a citation format.
Network · research()
Python and JavaScript SDK reference
Async patterns and Hybrid RAG.
Read · references/sdk.md
Framework integrations
LangChain, LlamaIndex, CrewAI, the Vercel AI SDK and other frameworks.
Read · references/integrations.md

From the official docs

Permissions

Low risk

The skill is documentation, but the code it helps you write calls Tavily's API with your key. Keep the key in an environment variable.

What it can do

Read your data
Create and change things
Delete things
Read-only mode

Watch out for

README's guided setup pipes an installer to bash
The tavily-ai/skills README recommends installing the Tavily CLI with `curl -fsSL https://cli.tavily.com/install.sh | bash`, then `tvly init`, which signs you in and installs the Tavily skills for the agents it detects. That route is separate from this skill; the install box here installs only the skill folder.
Shows pip and npm install commands
The Installation section lists pip install tavily-python and npm install @tavily/core. Claude may run them when you ask for an integration, so check the package names before you approve.
The code you build sends data to Tavily
Search queries and URLs go to Tavily's API using your key. Keep the key in an environment variable and out of source files.

Technical details

Talks to	Tavily API, from the code you build
Scopes	TAVILY_API_KEY (for code you write with it)

Questions

What does the tavily-best-practices skill do?

It gives Claude the Tavily SDK reference so it can write correct code for web search, content extraction, site crawling, URL discovery and research. It covers Python and JavaScript.

How do I install it in Claude Code, the Claude app or Codex?

Use the install box above. The Tavily README documents npx skills add https://github.com/tavily-ai/skills for other agents, and a guided tvly init setup that detects Claude Code, Codex and Cursor and installs the Tavily skills.

Does it need a Tavily API key?

The skill itself is documentation. The code it helps you write reads TAVILY_API_KEY from the environment, and you can get a key from Tavily. The SDK reference also shows passing the key directly.

Does it search the web for me?

No. It teaches Claude how to write Tavily code. For searching from the command line, the README lists tavily-search and other skills that use the Tavily CLI.

Is it safe to install?

The skill is Markdown only. Its Installation section shows pip install tavily-python and npm install @tavily/core, which Claude may run in your project. The code it writes sends queries to Tavily, so keep your API key in an environment variable.

Which frameworks does it cover?

The integrations guide covers LangChain, Pydantic AI, LlamaIndex, CrewAI, the Vercel AI SDK and other framework integrations.

How is it different from tavily-search?

This skill is SDK documentation for building integrations in code. The README describes tavily-search as the core skill that searches, extracts, maps and crawls through the Tavily CLI.

Information

updated Nov 2, 2026

Repository: tavily-ai/skills
Folder: skills/tavily-best-practices

tavily-ai/skills skills.sh

Sources

Docs, issue trackers and discussions this page is built from.

Show all 5

SKILL.md · github.com
tavily-ai/skills README · github.com
sdk.md · github.com
research.md · github.com
integrations.md · github.com