GrokSearch-rs MCP
An MCP server searches the web with Grok, retains sources and extracts page content through available retrieval providers.
- 453 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Queries and keys are sent to selected search services or a trusted server.
Install
- Works in
Manual install
npm install -g grok-search-rsThe first README Install command installs a native binary through npm.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Queries and keys are sent to selected search services or a trusted server.
This is third-party code. Review the repository files before installing.
What it does
GrokSearch-rs gives agents web search, source references and page retrieval. Search results stay cached so a client can request more sources separately without overflowing a single response. Specialist extractors handle sites such as GitHub, arXiv and Wikipedia. General pages use a configured Tavily, Exa, TinyFish and Firecrawl chain. Run locally over stdio or deploy HTTP access, supplying settings through the channel required by that transport.
Who it is for. For MCP-agent users combining search with source retrieval and bounded response sizes.
Good fit when
- Get a web answer with source links.
- Fetch additional sources in separate pages.
- Read a GitHub issue without a general scraper.
- Share search settings across MCP clients.
Not a fit when
- You need every page to work without provider setup.
- Queries cannot be sent to external search services.
- Model-returned links would be accepted without reading them.
Example request
Find sources for this question, show links and retrieve a selected page, identifying unavailable providers.MCP
- Transport
- stdio, http
- Authentication
- API key
Limitations
Grok search needs your own API access to the selected provider. Specialist extractors need no key, while ordinary URLs require at least one configured retrieval provider. A hosted server receives your keys in request headers, so use a trusted instance or local stdio. Full text remains cached even when an answer is short; account for private material and restrict access to the cache on the machine running the server.
How to disable. Remove the MCP entry and stop any self-hosted HTTP server. Clean cache, shared configuration and local tokens separately.
FAQ
Which main tools are available?
`web_search`, `get_sources`, `web_fetch`, `web_map` and `doctor`.
How do I create shared configuration?
Run `grok-search-rs --init`.
How do I retrieve more sources?
`get_sources` accepts `offset` and `limit` for cached pages.
Is Tavily mandatory?
No. The chain can use other supported providers.
Does xAI OAuth login have limitations?
It reuses the Hermes OAuth client ID. This may violate xAI terms or affect an account; an API key remains a separate connection option.
Discussion
No questions or reviews yet.
Related
MCP serversA web search MCP server that gives the agent full page text at once: Stack Overflow answers, GitHub threads, arXiv papers
A unified search MCP server: one interface to Tavily, Brave, Kagi, Exa, Linkup, Firecrawl and GitHub search
MCP server for web search and crawling: SearXNG and You.com plus clean Markdown from pages, sites and logged-in areas
Local web search for an agent with no API keys: results are ranked on your machine and only the relevant passages come back
Yandex and Google search MCP server
web-compass
An MCP web-search and content-extraction server via Yandex and Google on top of SerpAPI, geared toward RU/CIS results
Local web search for agents: searches, reads pages and returns compact passages with sources, saving context tokens
README in short
Search and extraction are separate. Failure of a generic provider does not make a specialist extractor depend on its key.