MCP SERVERWeb search through Tavily, tuned for models — search depth, news mode with a day window, domain filters, and an optional generated answer.
LABSCO SUMMARY
Raw HTML, images and image descriptions can each be larger than the results themselves, so having them opt-in keeps an ordinary search cheap and lets you turn them on for the one query that needs them. include_answer is the exception worth trying: a short generated answer alongside the sources gives the model something to check rather than something to trust. For news, prefer days over a general search — it is the difference between what is true now and what ranked well.
An MCP server over the Tavily Search API, which is built for retrieval rather than for humans reading a results page: it extracts the relevant content from each result and returns it in a form a model can use directly.
basic or advanced — and a result cap defaulting to 5 — tavily_searchtopic to news and days to how far back to look, defaulting to 3, or use time_range for a day, week, month or year windowinclude_domains and exclude_domains, so results come only from sources you acceptA Tavily API key — sign up on their site, a free tier exists — supplied as TAVILY_API_KEY, ideally in the client's env block rather than in a command. The npm package name is tavily-search-mcp-server (0.1.0); build from a clone with npm install and npm run build, or run the Docker image. Stdio is the default; running dist/sse.js, or setting TRANSPORT to sse in Docker with port 3001 published, serves it over SSE instead.
Build from source — git clone https://github.com/apappascs/tavily-search-mcp-server.git
USE CASES
Run a search mid-conversation and let the model read the extracted content rather than a snippet, for anything that changed after training.
Switch the topic to news and set the day window, so results are recent by construction instead of by luck.
Name the domains you accept, or exclude the ones you do not want quoted, and search normally otherwise.
Turn on raw content for the one query where you need the whole document, and leave it off the rest of the time.
Counted once with only startup setup. Other versions, modes or optional credentials may expose a different list.
tavily_searchPerforms a web search using the Tavily Search API, optimized for LLMs. Use this for broad information gathering, recent events, or when you need diverse web sources. Supports search depth, topic selection, time range filtering, and domain inclusion/exclusion.
BEFORE YOU INSTALL
1 tool when we connected. Extraction, images, generated answers, raw HTML, domain filters and time range are arguments to tavily_search, not separate tools.
TAVILY_API_KEY in your environment
Both documented setups run the server on your machine.
CONTEXT COST
Rated by the size of the one tool definition that rides along on every turn
INSTALLATION
You clone and build this one yourself.
git clone https://github.com/apappascs/tavily-search-mcp-server.git
cd tavily-search-mcp-server
npm install
npm run build{
"mcpServers": {
"tavily-search-server": {
"command": "node",
"args": [
"/Users/<YOUR_USERNAME>/<YOUR_FULL_PATH>/tavily-search-mcp-server/dist/index.js"
],
"env": {
"TAVILY_API_KEY": "<YOUR_TAVILY_API_KEY>"
}
}
}
}/Users/<YOUR_USERNAME>/<YOUR_FULL_PATH>/tavily-search-mcp-server is the full path of the folder you cloned (forward slashes, even on Windows), so the path ends in tavily-search-mcp-server/dist/index.js. Restart Claude Desktop afterwards.
{
"mcpServers": {
"tavily-search-server": {
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"-e",
"TAVILY_API_KEY",
"-v",
"/Users/<YOUR_USERNAME>/<YOUR_FULL_PATH>/tavily-search-mcp-server:/app",
"tavily-search-mcp-server"
],
"env": {
"TAVILY_API_KEY": "<YOUR_TAVILY_API_KEY>"
}
}
}
}Build the image first, inside your clone: docker build -t tavily-search-mcp-server:latest . — /Users/<YOUR_USERNAME>/<YOUR_FULL_PATH>/tavily-search-mcp-server is the full path of that clone (forward slashes, even on Windows). Restart Claude Desktop afterwards.
MCP servers
mrkrsl
Tomatio13
baranwang