MCP servers
Official MCP servers for web search and scraping
Give agents fresh information from the web: search, page fetching, crawling and structured extraction.
Reviewed Oct 5, 2026. Every server here is published under the vendor's own namespace in the official MCP registry.
| Server | What it does | Runs | Version |
|---|---|---|---|
| ApifyApify | Run thousands of ready-made scrapers and automations from Apify Store as tools. | Remote | 0.17.2 |
| Brave SearchBrave | Web, image, video and news results plus AI summaries from Brave's independent index. | Local | 2.1.3 |
| ExaExa | Neural web search, page fetching and multi-step research for agents. | Remote | 3.4.1 |
| FirecrawlFirecrawl | Search the web and scrape or crawl pages into clean Markdown for the model. | Remote + local | 3.27.3 |
| PerplexityPerplexity | Real-time web search and research through the Perplexity API. | Local | 1.2.1 |
| TavilyTavily | Real-time web search, extraction and crawl tuned for agents. | Local | 0.2.15 |
What teams use these for
- Research a topic with Exa, Tavily, Brave or Perplexity search
- Turn a docs site into Markdown with Firecrawl
- Run an Apify scraper for structured data
How to choose
- Remote first. A hosted server with OAuth means no API key on your laptop and updates without reinstalling.
- Local when data must stay put. Packages run on your machine, which suits private networks and self-hosted instances.
- Fewer tools, better results. Every connected server adds tool descriptions to the context. Connect what the task needs, not everything.
Other categories
Built an MCP server for web search & scraping?
Feature it at the top of this page and the directory from $49/month, or list it for free.