Skip to main content
Glama
jackedelic

tavily-mcp

by jackedelic
OverviewSchemaSourceRelated ServersDiscussionsAdmin
JavaScript
Remote

Tavily Crawl Beta

GitHub Repo stars npm

🎉 Introducing tavily-crawl + tavily-map in v0.2.1! 🎉

MCP demo

The Model Context Protocol (MCP) is an open standard that enables AI systems to interact seamlessly with various data sources and tools, facilitating secure, two-way connections.

Developed by Anthropic, the Model Context Protocol (MCP) enables AI assistants like Claude to seamlessly integrate with Tavily's advanced search and data extraction capabilities. This integration provides AI models with real-time access to web information, complete with sophisticated filtering options and domain-specific search features.

The Tavily MCP server provides:

  • search, extract, map, crawl tools

  • Real-time web search capabilities through the tavily-search tool

  • Intelligent data extraction from web pages via the tavily-extract tool

  • Powerful web mapping tool that creates a structured map of website

  • Web crawler that systematically explores websites

📚 Helpful Resources

  • Tutorial on combining Tavily MCP with Neo4j MCP server

  • Tutorial on integrating Tavily MCP with Cline in VS Code

Related MCP server: tavily-mcp

Prerequisites 🔧

Before you begin, ensure you have:

  • Tavily API key

    • If you don't have a Tavily API key, you can sign up for a free account here

  • Claude Desktop or Cursor

  • Node.js (v20 or higher)

    • You can verify your Node.js installation by running:

      • node --version

  • Git installed (only needed if using Git installation method)

    • On macOS: brew install git

    • On Linux:

      • Debian/Ubuntu: sudo apt install git

      • RedHat/CentOS: sudo yum install git

    • On Windows: Download Git for Windows

Tavily MCP server installation ⚡

Running with NPX

npx -y tavily-mcp@0.2.1

Installing via Smithery

To install Tavily MCP Server for Claude Desktop automatically via Smithery:

npx -y @smithery/cli install @tavily-ai/tavily-mcp --client claude

Although you can launch a server on its own, it's not particularly helpful in isolation. Instead, you should integrate it into an MCP client. Below is an example of how to configure the Claude Desktop app to work with the tavily-mcp server.

Configuring MCP Clients ⚙️

This repository will explain how to configure VS Code, Cursor and Claude Desktop to work with the tavily-mcp server.

Configuring VS Code 💻

For one-click installation, click one of the install buttons below:

Install with NPX in VS Code Install with NPX in VS Code Insiders

Manual Installation

First check if there are install buttons at the top of this section that match your needs. If you prefer manual installation, follow these steps:

Add the following JSON block to your User Settings (JSON) file in VS Code. You can do this by pressing Ctrl + Shift + P (or Cmd + Shift + P on macOS) and typing Preferences: Open User Settings (JSON).

{
  "mcp": {
    "inputs": [
      {
        "type": "promptString",
        "id": "tavily_api_key",
        "description": "Tavily API Key",
        "password": true
      }
    ],
    "servers": {
      "tavily": {
        "command": "npx",
        "args": ["-y", "tavily-mcp@0.2.1"],
        "env": {
          "TAVILY_API_KEY": "${input:tavily_api_key}"
        }
      }
    }
  }
}

Optionally, you can add it to a file called .vscode/mcp.json in your workspace:

{
  "inputs": [
    {
      "type": "promptString",
      "id": "tavily_api_key",
      "description": "Tavily API Key",
      "password": true
    }
  ],
  "servers": {
    "tavily": {
      "command": "npx",
      "args": ["-y", "tavily-mcp@0.2.1"],
      "env": {
        "TAVILY_API_KEY": "${input:tavily_api_key}"
      }
    }
  }
}

Configuring Cline 🤖

The easiest way to set up the Tavily MCP server in Cline is through the marketplace with a single click:

  1. Open Cline in VS Code

  2. Click on the Cline icon in the sidebar

  3. Navigate to the "MCP Servers" tab ( 4 squares )

  4. Search "Tavily" and click "install"

  5. When prompted, enter your Tavily API key

Alternatively, you can manually set up the Tavily MCP server in Cline:

  1. Open the Cline MCP settings file:

    For macOS:

    # Using Visual Studio Code
code ~/Library/Application\ Support/Code/User/globalStorage/saoudrizwan.claude-dev/settings/cline_mcp_settings.json

# Or using TextEdit
open -e ~/Library/Application\ Support/Code/User/globalStorage/saoudrizwan.claude-dev/settings/cline_mcp_settings.json

    For Windows:

    code %APPDATA%\Code\User\globalStorage\saoudrizwan.claude-dev\settings\cline_mcp_settings.json

  2. Add the Tavily server configuration to the file:

    Replace your-api-key-here with your actual Tavily API key.

    {
  "mcpServers": {
    "tavily-mcp": {
      "command": "npx",
      "args": ["-y", "tavily-mcp@0.2.1"],
      "env": {
        "TAVILY_API_KEY": "your-api-key-here"
      },
      "disabled": false,
      "autoApprove": []
    }
  }
}

  3. Save the file and restart Cline if it's already running.

  4. When using Cline, you'll now have access to the Tavily MCP tools. You can ask Cline to use the tavily-search and tavily-extract tools directly in your conversations.

Configuring Cursor 🖥️

Note: Requires Cursor version 0.45.6 or higher

To set up the Tavily MCP server in Cursor:

  1. Open Cursor Settings

  2. Navigate to Features > MCP Servers

  3. Click on the "+ Add New MCP Server" button

  4. Fill out the following information:

    • Name: Enter a nickname for the server (e.g., "tavily-mcp")

    • Type: Select "command" as the type

    • Command: Enter the command to run the server:

      env TAVILY_API_KEY=your-api-key npx -y tavily-mcp@0.2.1

      Important: Replace your-api-key with your Tavily API key. You can get one at app.tavily.com/home

After adding the server, it should appear in the list of MCP servers. You may need to manually press the refresh button in the top right corner of the MCP server to populate the tool list.

The Composer Agent will automatically use the Tavily MCP tools when relevant to your queries. It is better to explicitly request to use the tools by describing what you want to do (e.g., "User tavily-search to search the web for the latest news on AI"). On mac press command + L to open the chat, select the composer option at the top of the screen, beside the submit button select agent and submit the query when ready.

Cursor Interface Example

Configuring the Claude Desktop app 🖥️

For macOS:

# Create the config file if it doesn't exist
touch "$HOME/Library/Application Support/Claude/claude_desktop_config.json"

# Opens the config file in TextEdit 
open -e "$HOME/Library/Application Support/Claude/claude_desktop_config.json"

# Alternative method using Visual Studio Code (requires VS Code to be installed)
code "$HOME/Library/Application Support/Claude/claude_desktop_config.json"

For Windows:

code %APPDATA%\Claude\claude_desktop_config.json

Add the Tavily server configuration:

Replace your-api-key-here with your actual Tavily API key.

{
  "mcpServers": {
    "tavily-mcp": {
      "command": "npx",
      "args": ["-y", "tavily-mcp@0.2.1"],
      "env": {
        "TAVILY_API_KEY": "your-api-key-here"
      }
    }
  }
}

2. Git Installation

  1. Clone the repository:

git clone https://github.com/tavily-ai/tavily-mcp.git
cd tavily-mcp

  1. Install dependencies:

npm install

  1. Build the project:

npm run build

Configuring the Claude Desktop app ⚙️

Follow the configuration steps outlined in the Configuring the Claude Desktop app section above, using the below JSON configuration.

Replace your-api-key-here with your actual Tavily API key and /path/to/tavily-mcp with the actual path where you cloned the repository on your system.

{
  "mcpServers": {
    "tavily": {
      "command": "npx",
      "args": ["/path/to/tavily-mcp/build/index.js"],
      "env": {
        "TAVILY_API_KEY": "your-api-key-here"
      }
    }
  }
}

Usage in Claude Desktop App 🎯

Once the installation is complete, and the Claude desktop app is configured, you must completely close and re-open the Claude desktop app to see the tavily-mcp server. You should see a hammer icon in the bottom left of the app, indicating available MCP tools, you can click on the hammer icon to see more detial on the tavily-search and tavily-extract tools.

Alt text

Now claude will have complete access to the tavily-mcp server, including the tavily-search and tavily-extract tools. If you insert the below examples into the Claude desktop app, you should see the tavily-mcp server tools in action.

Tavily Search Examples

  1. General Web Search:

Can you search for recent developments in quantum computing?

  1. News Search:

Search for news articles about AI startups from the last 7 days.

  1. Domain-Specific Search:

Search for climate change research on nature.com and sciencedirect.com

Tavily Extract Examples

  1. Extract Article Content:

Extract the main content from this article: https://example.com/article

✨ Combine Search and Extract ✨

You can also combine the tavily-search and tavily-extract tools to perform more complex tasks.

Search for news articles about AI startups from the last 7 days and extract the main content from each article to generate a detailed report.

Troubleshooting 🛠️

Common Issues

  1. Server Not Found

    • Verify the npm installation by running npm --verison

    • Check Claude Desktop configuration syntax by running code ~/Library/Application\ Support/Claude/claude_desktop_config.json

    • Ensure Node.js is properly installed by running node --version

  2. NPX related issues

  • If you encounter errors related to npx, you may need to use the full path to the npx executable instead.

  • You can find this path by running which npx in your terminal, then replace the "command": "npx" line with "command": "/full/path/to/npx" in your configuration.

  1. API Key Issues

    • Confirm your Tavily API key is valid

    • Check the API key is correctly set in the config

    • Verify no spaces or quotes around the API key

Acknowledgments ✨

Available Tools

4 tools
tavily-crawlA

A powerful web crawler that initiates a structured web crawl starting from a specified base URL. The crawler expands from that point like a tree, following internal links across pages. You can control how deep and wide it goes, and guide it to focus on specific sections of the site.

ParametersJSON Schema
NameRequiredDescriptionDefault
urlYesThe root URL to begin the crawl
max_depthNoMax depth of the crawl. Defines how far from the base URL the crawler can explore.
max_breadthNoMax number of links to follow per level of the tree (i.e., per page)
limitNoTotal number of links the crawler will process before stopping
instructionsNoNatural language instructions for the crawler
select_pathsNoRegex patterns to select only URLs with specific path patterns (e.g., /docs/.*, /api/v1.*)
select_domainsNoRegex patterns to select crawling to specific domains or subdomains (e.g., ^docs\.example\.com$)
allow_externalNoWhether to allow following links that go to external domains
categoriesNoFilter URLs using predefined categories like documentation, blog, api, etc
extract_depthNoAdvanced extraction retrieves more data, including tables and embedded content, with higher success but may increase latencybasic

TDQS

A3.9/5.0
Behavior3/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

With no annotations provided, the description carries the full burden of behavioral disclosure. It effectively describes the crawling behavior ('expands like a tree', 'following internal links'), scope control ('how deep and wide it goes'), and focus guidance. However, it doesn't mention important behavioral aspects like rate limits, authentication requirements, error handling, or what the output format looks like.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is perfectly sized at three sentences, each earning its place. It's front-loaded with the core purpose, followed by expansion behavior, and ending with control capabilities. Zero wasted words or redundant information.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness3/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given the complexity of a 10-parameter web crawling tool with no annotations and no output schema, the description provides adequate high-level context but lacks details about output format, error conditions, performance characteristics, or specific use cases. For a tool this complex, more behavioral and output information would be helpful despite the excellent schema coverage.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 100%, so the schema already documents all 10 parameters thoroughly. The description adds some high-level context about controlling depth, breadth, and focusing on specific sections, which aligns with parameters like max_depth, max_breadth, and categories/select_paths. However, it doesn't provide additional semantic meaning beyond what's already in the comprehensive schema descriptions.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states the tool's purpose with specific verbs ('initiates a structured web crawl', 'expands like a tree', 'following internal links') and identifies the resource ('starting from a specified base URL'). It distinguishes this crawl tool from sibling tools like 'search' or 'extract' by emphasizing its tree-based expansion approach.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description provides clear context about when to use this tool: for structured crawling starting from a base URL with tree-like expansion. It mentions controlling depth, breadth, and focusing on specific site sections. However, it doesn't explicitly state when NOT to use it or name specific alternatives among the sibling tools.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

tavily-extractC

A powerful web content extraction tool that retrieves and processes raw content from specified URLs, ideal for data collection, content analysis, and research tasks.

ParametersJSON Schema
NameRequiredDescriptionDefault
urlsYesList of URLs to extract content from
extract_depthNoDepth of extraction - 'basic' or 'advanced', if usrls are linkedin use 'advanced' or if explicitly told to use advancedbasic
include_imagesNoInclude a list of images extracted from the urls in the response

TDQS

C2.9/5.0
Behavior2/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

With no annotations provided, the description carries the full burden of behavioral disclosure. It mentions 'retrieves and processes raw content' but lacks details on rate limits, authentication needs, error handling, or what 'processes' entails (e.g., formatting, cleaning). For a web extraction tool with potential complexities, this leaves significant gaps in understanding its behavior beyond basic functionality.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is a single, efficient sentence that front-loads the core purpose ('retrieves and processes raw content from specified URLs') and adds value with ideal use cases. There's no wasted wording, though it could be slightly more structured by separating functional description from usage contexts.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness2/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given the tool's complexity (web content extraction with processing), lack of annotations, and no output schema, the description is incomplete. It doesn't explain what 'processes' means, the format or structure of returned content, potential limitations (e.g., site restrictions), or how it differs from siblings. For a tool with three parameters and no structured behavioral hints, more context is needed.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

The input schema has 100% description coverage, clearly documenting all three parameters. The description adds no parameter-specific information beyond what the schema provides, such as examples or contextual usage tips. However, since the schema is comprehensive, a baseline score of 3 is appropriate as the description doesn't need to compensate for gaps.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose4/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states the tool's purpose as 'retrieves and processes raw content from specified URLs' with specific verbs and resources. It distinguishes itself from potential siblings by focusing on extraction rather than crawling, mapping, or searching, though it doesn't explicitly name alternatives. The description is specific but could be more precise about what 'processes' entails.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines2/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description provides minimal guidance with 'ideal for data collection, content analysis, and research tasks,' but offers no explicit when-to-use rules, exclusions, or comparisons to sibling tools like tavily-crawl, tavily-map, or tavily-search. There's no mention of prerequisites, limitations, or scenarios where this tool is preferred over alternatives.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

tavily-mapA

A powerful web mapping tool that creates a structured map of website URLs, allowing you to discover and analyze site structure, content organization, and navigation paths. Perfect for site audits, content discovery, and understanding website architecture.

ParametersJSON Schema
NameRequiredDescriptionDefault
urlYesThe root URL to begin the mapping
max_depthNoMax depth of the mapping. Defines how far from the base URL the crawler can explore
max_breadthNoMax number of links to follow per level of the tree (i.e., per page)
limitNoTotal number of links the crawler will process before stopping
instructionsNoNatural language instructions for the crawler
select_pathsNoRegex patterns to select only URLs with specific path patterns (e.g., /docs/.*, /api/v1.*)
select_domainsNoRegex patterns to select crawling to specific domains or subdomains (e.g., ^docs\.example\.com$)
allow_externalNoWhether to allow following links that go to external domains
categoriesNoFilter URLs using predefined categories like documentation, blog, api, etc

TDQS

A3.9/5.0
Behavior3/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

No annotations are provided, so the description carries the full burden. It describes the tool as 'powerful' and for 'discovering and analyzing,' which implies it performs read-only operations, but doesn't specify behavioral traits like rate limits, authentication needs, or potential impacts on target websites. The description adds value by explaining the mapping purpose but lacks detailed behavioral disclosure.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is appropriately sized and front-loaded: it starts with the core purpose, then elaborates on use cases. Every sentence earns its place by adding value without redundancy, making it efficient and well-structured.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness3/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given the complexity (9 parameters, no annotations, no output schema), the description is somewhat complete but has gaps. It explains the tool's purpose and use cases well, but without annotations or output schema, it doesn't cover behavioral aspects or return values, leaving the agent to infer details from the schema alone.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

The schema description coverage is 100%, so the schema already documents all 9 parameters thoroughly. The description doesn't add specific parameter semantics beyond what the schema provides, such as explaining how 'categories' interact with mapping or the implications of 'max_depth.' Baseline 3 is appropriate since the schema does the heavy lifting.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states the tool's purpose: 'creates a structured map of website URLs' with specific verbs like 'discover and analyze site structure, content organization, and navigation paths.' It distinguishes from siblings like 'search' or 'extract' by focusing on mapping and structural analysis rather than general search or content extraction.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description provides clear context for when to use this tool: 'Perfect for site audits, content discovery, and understanding website architecture.' It doesn't explicitly mention when not to use it or name alternatives among siblings, but the context strongly implies it's for structural mapping rather than other tasks like searching or crawling without mapping.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

  1. 4 tool updates
    • First observedtavily-crawl
    • First observedtavily-extract
    • First observedtavily-map
    • First observedtavily-search
Deploy Server
One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

TDQS

A3.6/5.0

Scored across 4 tools

Disambiguation4/5

The tools have mostly distinct purposes: crawling (tavily-crawl), extraction (tavily-extract), mapping (tavily-map), and searching (tavily-search). However, there is some potential overlap between tavily-crawl and tavily-map, as both involve exploring website structure, which could cause minor confusion for an agent.

Naming Consistency5/5

All tool names follow a consistent 'tavily-' prefix with a descriptive action suffix (crawl, extract, map, search). This pattern is clear and predictable, making it easy for agents to understand and use the toolset.

Tool Count5/5

With 4 tools, this server is well-scoped for web-related tasks. Each tool serves a specific function in the domain of web data retrieval and analysis, and none appear redundant or unnecessary given the server's purpose.

Completeness4/5

The toolset covers key web operations: crawling, extracting, mapping, and searching. Minor gaps might include advanced filtering or processing options, but the core workflows for web content analysis and research are well-covered, allowing agents to perform most common tasks effectively.

Maintenance

ActivityInactive
ResponsivenessNo issues

Resources

Related MCP Connectors

Related MCP Servers

Appeared in Searches