Compared · Web fetching & scraping
Best MCP servers for Web fetching & scraping
5 entries in the AgentTask Atlas catalog fit web fetching & scraping. They are not ranked by a score: the table lists active entries first, then by how many clients document a setup, and each row ends with our one-line verdict on when to pick it. Official sources last checked 2026-10-21.
- No credentials needed: Playwright MCP, Fetch
- Offers a hosted/remote option: Exa MCP Server, Tavily MCP Server
- Most documented clients: Cursor (4), VS Code (4), Claude Code (3), Windsurf (3), Claude Desktop (3), Codex (2).
| Name | Type | Credentials | License | Clients | Verdict |
|---|---|---|---|---|---|
| Exa MCP Server Exa | MCP server | Optional | MIT | 12 | Pick it for web search and clean page fetching in the agent without an API key to start; add OAuth or a key for higher limits and multi-step Exa Agent research. |
| Playwright MCP Microsoft | MCP server | None | Apache-2.0 | 10 | Pick it to let the agent drive a real browser for E2E checks; it can reach any site the browser can. |
| Firecrawl MCP Server Firecrawl | MCP server | Required | MIT | 4 | Pick it for crawling and scraping at scale; URLs are sent to Firecrawl's hosted API with your key. |
| Tavily MCP Server Tavily | MCP server | Required | MIT | 3 | Pick it for real-time web search plus page extraction, site mapping and crawling in one server; needs a Tavily key or OAuth, and every query goes to Tavily. |
| Fetch MCP steering group (reference) | MCP server | None | NOASSERTION | 1 | Pick it to pull a web page into context; the README warns it can reach local/internal IPs, so restrict it. |
Install notes
- Exa MCP Server
- Connect to https://mcp.exa.ai/mcp. Claude Code: `claude plugin install exa@claude-plugins-official`; Codex: `codex mcp add exa --url https://mcp.exa.ai/mcp` (see README for other clients). official docs ↗Credentials: Optional: works anonymously with rate limits; Exa OAuth or an Exa API key for higher limits and Exa Agent. · Maintenance: Hosted remote server (MCP Registry ai.exa/exa v3.4.1); no GitHub release; last commit 2026-10-19
- Playwright MCP
- Add `npx @playwright/mcp@latest` to your client MCP config (README has per-client steps). official docs ↗Credentials: None · Maintenance: Release v0.0.83 on 2026-09-28; last commit 2026-10-08
- Firecrawl MCP Server
- `npx -y firecrawl-mcp` with FIRECRAWL_API_KEY (see README). official docs ↗Credentials: FIRECRAWL_API_KEY. · Maintenance: Release v3.2.1 on 2025-09-26; last commit 2026-10-16
- Tavily MCP Server
- Remote: connect to https://mcp.tavily.com/mcp. Claude Code: `claude mcp add --transport http tavily https://mcp.tavily.com/mcp` (OAuth) or append `?tavilyApiKey=<your-api-key>`; local: `npx -y tavily-mcp@latest` with TAVILY_API_KEY. official docs ↗Credentials: Tavily API key (free tier available) in the URL or an Authorization header, or OAuth on the remote server. · Maintenance: npm tavily-mcp v0.2.22; no GitHub release; last commit 2026-10-15
- Fetch
- Run with `uvx mcp-server-fetch` (see README). official docs ↗Credentials: None · Maintenance: Monorepo release v2.0.0 on 2026-10-19 publishes mcp-server-fetch 2026.10.19 (PyPI); last commit 2026-10-19 (path-scoped)
Set up any of these in your client: Claude Code · Cursor · VS Code · Claude Desktop. Filter interactively on the Web fetching & scraping explorer, or ask the catalog from your agent with the free MCP server.
Other comparisons: Databases & SQL · Infra & deploy · Debugging & observability · Library docs & research · Frontend & UI · Local files · Docs & writing
⚠ No independent security audit: entries summarize official sources only. Review permissions and code before installing.