MCP servers for a web research agent
The MCP server bundle to build an AI that researches topics on the live web and cites sources.
3 servers · ~18 tools · Last updated June 17, 2026
TL;DR: Research agents need both discovery (search) and depth (fetch and clean the actual pages). This bundle combines a semantic search engine with scrapers so the agent finds the right sources and reads them in full.
Bottom line: start with Exa Search and add the rest as your needs grow. All 3 install together via the merged config below (~18 tools total).
Tool budget: this stack exposes about 18 tools. That's within Cursor's practical ~40-tool ceiling, so all servers can stay enabled together. Check your own config →
What's in the stack
One-click config
All 3 servers merged into a single block — pick your client and paste.
{
"mcpServers": {
"exa-search": {
"command": "npx",
"args": [
"-y",
"exa-mcp-server"
],
"env": {
"EXA_API_KEY": "<your-exa-api-key>"
}
},
"firecrawl": {
"command": "npx",
"args": [
"-y",
"firecrawl-mcp"
],
"env": {
"FIRECRAWL_API_KEY": "<your-firecrawl-api-key>"
}
},
"tavily": {
"command": "npx",
"args": [
"-y",
"tavily-mcp@latest"
],
"env": {
"TAVILY_API_KEY": "<your-tavily-api-key>"
}
}
}
}Paste into ~/Library/Application Support/Claude/claude_desktop_config.json and fully restart Claude Desktop. Replace any placeholder keys/tokens with your own.
Capabilities this stack covers
FAQ
Why two search/scrape servers?
Search finds candidate sources; scraping reads them in full. Different engines also have different coverage, so combining improves recall.
Will this cite sources?
These servers return URLs with results, so the agent can attribute claims — just prompt it to include citations.