Tools
Tavily
Follow these instructions to quickly run the Tavily MCP server and enable its tools inside a Meshagent Room:
Required Environment Variables
TAVILY_API_KEY: Your Tavily API key.
Steps
-
Install the Meshagent CLI and Agents
-
Sign up and authenticate
- Register and log in following the CLI get started guide.
-
Launch the Tavily MCP Agent as a Service
- This command creates a Meshagent room called
testwith the Tavily MCP server available as a callable service in the room. - Note: Meshagent rooms will close automatically if they go inactive.
- This command creates a Meshagent room called
-
Join the Room with a Chatbot and Tools
- This will start a chatbot in the Meshagent room, able to use the Tavily toolkit.
- Multiple toolkits can operate in the same room by repeating this command with different configurations.
- After launching, you’ll receive a link in the command output to access the room, chat, and use the tools interactively.
- This will start a chatbot in the Meshagent room, able to use the Tavily toolkit.
Tools Available
Tools Provided by This Server
| Tool Name | Short Description |
|---|---|
tavily-crawl | A web crawler that initiates a structured crawl from a specified base URL. |
tavily-extract | Retrieves and processes raw content from URLs for data collection and content analysis. |
tavily-map | Creates a structured map of site URLs to discover content organization and navigation paths. |
tavily-search | Real-time web search with Tavily’s AI search engine for comprehensive, current results. |
Tool Details
tavily-crawl
Initiates a structured crawl from a base URL, expanding tree-like across internal links. Supports filters, crawl depth, and custom instructions.
| Parameter | Type | Description |
|---|---|---|
url | string | Root URL to begin crawl |
allow_external | boolean | Optional. Follow links to external domains |
categories | array | Optional. Filter by categories (documentation, blog, api, etc.) |
extract_depth | string | Optional. Advanced extraction expands data (increases latency) |
instructions | string | Optional. Natural language instructions for the crawler |
limit | integer | Optional. Max links to process |
max_breadth | integer | Optional. Max links to follow per level |
max_depth | integer | Optional. Max tree depth |
select_domains | array | Optional. Regex patterns for included domains |
select_paths | array | Optional. Regex patterns for included paths |
tavily-extract
Extracts and processes content from given URLs, including deep extraction modes and images.
| Parameter | Type | Description |
|---|---|---|
urls | array | List of URLs to extract content from |
extract_depth | string | Optional. ‘basic’ or ‘advanced’ extraction |
include_images | boolean | Optional. Return images extracted from the URLs |
tavily-map
Builds a map of site URLs, showing structure and navigation.
| Parameter | Type | Description |
|---|---|---|
url | string | Root URL for mapping |
allow_external | boolean | Optional. Follow external links |
categories | array | Optional. Filter URLs by categories |
instructions | string | Optional. Natural language crawler instructions |
limit | integer | Optional. Max links to process |
max_breadth | integer | Optional. Max links per level |
max_depth | integer | Optional. Depth limit |
select_domains | array | Optional. Regex for domain inclusion |
select_paths | array | Optional. Regex for path inclusion |
tavily-search
Performs real-time web search using Tavily’s AI engine.
| Parameter | Type | Description |
|---|---|---|
query | string | Search query |
days | number | Optional. Number of days back (for ‘news’ topic searches) |
exclude_domains | array | Optional. Domains to exclude |
include_domains | array | Optional. Domains to include |
include_image_descriptions | boolean | Optional. Include images + their descriptions |
include_images | boolean | Optional. Include related images |
include_raw_content | boolean | Optional. Include cleaned HTML content |
max_results | number | Optional. Maximum number of results |
search_depth | string | Optional. ‘basic’ or ‘advanced’ depth |
time_range | string | Optional. Time window for search |
topic | string | Optional. Category/topic of search |