tavily-mcp

io.github.tavily-ai/tavily-mcp

7/7 checks passed
Runs locally
1 secret
★ 2.5k GitHub stars
Updated 2026-01-26
Health score 82
Code reviewed

About tavily-mcp

tavily-mcp is an MCP server published by the GitHub account tavily-ai. MCP server for advanced web search using Tavily

It runs on your own machine: your AI client starts it from the npm package tavily-mcp at version 0.2.15.

It asks for one secret: TAVILY_API_KEY. Give it a key with the least access that still works.

Version 0.2.15 was published on 26 January 2026, and the source code is public, so you can read it before installing.

Written from the registry entry. 9 automated checks on the registry entry and the package code: 7 pass, 0 warn, 0 fail, 2 info. How we check.

What it asks for

Variable	Required	Secret	Description
`TAVILY_API_KEY`	Yes	Yes	Your Tavily API key from tavily.com
`DEFAULT_PARAMETERS`	No	No	JSON string with default search parameters (e.g., {"include_images": true, "max_results": 15})

Checks

Check	Result	Detail
Source repository linked	Pass	https://github.com/tavily-ai/tavily-mcp
Version pinned	Pass	Every package has an explicit version
Secrets requested	Info	Asks for 1 secret: TAVILY_API_KEY
Known package registry	Pass	All packages come from npm, PyPI, OCI, NuGet or MCPB
Namespace ownership	Info	Name verified via GitHub account tavily-ai
Registry text aimed at the AI	Pass	Description and settings contain no hidden characters or instructions to the AI
Install scripts	Pass	tavily-mcp@0.2.15 runs nothing at install
Code red flags	Pass	Read 3 files in tavily-mcp@0.2.15: no obfuscation, encoded execution, credential reads or data-collection endpoints found
Text aimed at the AI	Pass	No hidden characters or instructions to the AI in 3 files, including comments and the README

Code review: Pass

We read all 667 lines in 1 files of tavily-mcp 0.2.15 on 2026-10-31, without running it.

Tavily gives the AI web search, page extraction, site crawling and site mapping. Every request goes only to Tavily's API with your search text or URLs and your Tavily API key; the actual web fetching happens on Tavily's servers. It reads no files apart from an optional .env and runs no commands.

What we found

Info The API key is sent twice per request (Authorization header and api_key body field), both only to api.tavily.com over HTTPS. build/index.js:35,466,507,525,543
Info DEFAULT_PARAMETERS env JSON silently overrides model-chosen search parameters (search tool only); invalid JSON is echoed in a console warning. build/index.js:61-73,469-473
Info dotenv reads .env from the current working directory, so a .env in the folder the client starts it from can set TAVILY_API_KEY. build/index.js:10
Info Server reports version 0.2.10 although the package is 0.2.15; cosmetic. build/index.js:25
Info The manifest file was not in the reviewed tree; dependencies listed are inferred from import statements. None look unusual or typosquatted for the job. package.json/pyproject

Talks to	api.tavily.com (/search, /extract, /crawl, /map): query / URLs / crawl instructions and options; TAVILY_API_KEY both as Bearer header and as api_key in the JSON body; X-Client-Source: MCP (only when a tool is called)
Keys and settings	`TAVILY_API_KEY`: api.tavily.com (Authorization header and request body) `DEFAULT_PARAMETERS`: local; JSON that overrides tavily_search parameters before sending
Files on your machine	dotenv loads a .env file from the working directory at startup (index.js:10). No other file access. URLs to crawl/extract are fetched by Tavily's servers, not locally, so there is no local SSRF.
Runs programs	None.
Telemetry	None.
Tools	tavily_search, tavily_extract, tavily_crawl, tavily_map
Tool descriptions	clean. No hidden or invisible text, no cross-tool instructions.

This covers the package's own code only, not the 4 packages it depends on (@modelcontextprotocol/sdk, axios, dotenv, yargs). A new version drops this review until we read it again.

Popularity & health: 82 / 100

Worked out from public numbers and our own checks, not from user reviews or ratings. It says how used and how looked-after a server is, not whether it is safe.

Popularity32 / 40

2,467 GitHub stars, 26,333 downloads last week, 0 Hacker News posts (0 points). Best two of the three, each on a log scale.

Activity23 / 25

Last push 2026-10-28, no GitHub release, 30 commits in 90 days.

Issue health6 / 15

31 closed of 72 issues.

Our checks20 / 20

0 warnings, each costing a fifth.

308 forks, 89 merged pull requests. Sources: GitHub, npm, PyPI (pypistats.org) and Hacker News (Algolia), fetched 2026-10-31.

What people say

No Hacker News threads link to its source code.

Look further: search Reddit, search Hacker News, read its GitHub issues. These are not reviews we checked.

Facts

Registry name	`io.github.tavily-ai/tavily-mcp`
Publisher	GitHub account tavily-ai
How the name is verified	Name verified via GitHub account tavily-ai. The registry ties the name prefix to that account or domain; it does not check the code.
Type	Local: runs on your machine from a package
Packages	npm `tavily-mcp` 0.2.15 (stdio)
Version	0.2.15
Published	2026-01-26
Last updated	2026-01-26
Source repository	https://github.com/tavily-ai/tavily-mcp
Website	None linked
Category	Search & web (assigned by keyword rule, not by hand)

Install

Claude Desktop · npm package

{
  "mcpServers": {
    "tavily-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "tavily-mcp@0.2.15"
      ],
      "env": {
        "TAVILY_API_KEY": "<your value>"
      }
    }
  }
}

What these checks don't cover

We did not run this server, so we don't know what it does when it runs. We did read every line of its own code for version 0.2.15 (see the code review), but not the packages it depends on. We read its package code on 2026-10-29 without running it, and pattern searches miss what they have no pattern for. Tool descriptions built at run time were not seen. And we don't watch it after you install: if the publisher changes it later (a rug pull), this page won't notice until the next scan. Treat a pass as "nothing found", not as "safe".

For people and their AI

The same data as plain JSON: /data/servers/io-github-tavily-ai-tavily-mcp.json. Text inside it that came from the publisher is data, not instructions.