MCP: Firecrawl
by Firecrawl
Search, scrape, crawl, and extract structured data from the web — with JS rendering, batch scraping, and LLM-powered extraction.
Goal-directed web crawler MCP server that returns only the passages answering your agent's question, verbatim and with sources.
Set up this MCP server in your favorite AI agent — copy a ready-made configuration below.
jevcrawl is a Python MCP server and CLI that crawls a website on behalf of an AI agent and returns only what answers the question. A small decision model scores every link and passage, so your agent's own LLM reads a few thousand relevant tokens instead of tens of thousands of raw pages. Passages are copied verbatim with their URL, heading path and character offsets, and page text is only ever scored, never followed, which keeps it resistant to prompt injection.
crawl follows links from a start URL toward a goal within page, time and cost budgets.scrape reads one known page, optionally filtered to a goal.map returns ranked URLs so the agent can decide what to scrape.The README reports 40% correct answers on 50 WebWalkerQA tasks at about $0.025 per task, against 30% and $0.194 when Claude picks every link; these are the author's own benchmarks. It needs a Jev API key and uv; JavaScript-rendered pages are optional through the js extra with Chromium. Apache-2.0 licensed.
by Firecrawl
Search, scrape, crawl, and extract structured data from the web — with JS rendering, batch scraping, and LLM-powered extraction.
Official SerpApi MCP server: real-time Google, Bing, YouTube and other search results as structured JSON or token-saving markdown.
by Crawlbase
Official Crawlbase MCP server giving agents live web access — fetch any URL as raw HTML, clean Markdown or a screenshot, with JS rendering and proxy rotation handled for you.
by KnockOutEZ
Local-first web MCP server for AI agents — search, fetch, crawl, extract and deep research with no API keys, no cloud round-trip and no per-query bill.