Puppeteer MCP Server vs Fetch MCP Server
When Claude needs a web page, Fetch grabs the HTML and converts it to markdown — fast, cheap, and still actively maintained. Puppeteer drives a real browser: JavaScript-rendered pages, screenshots, form interaction — but the reference server is archived. Rule of thumb: start with Fetch, and only take on a browser's weight when the page genuinely requires one.
GitHub data verified 2026-07-30 — objective fields only, no editorial scoring.
| Puppeteer MCP Server Browser automation and web scraping with headless Chrome | Fetch MCP Server Make HTTP requests to any URL with configurable options | |
|---|---|---|
| GitHub stars | 293 | 89,034 |
| Forks | 165 | 11,331 |
| Open issues | 0 | 479 |
| Last commit | 2025-05-28 | 2026-07-29 |
| Maintenance status | Archived (read-only) | Actively maintained |
| License | MIT | Other |
| Language | TypeScript | TypeScript |
| Install | npx @modelcontextprotocol/server-puppeteer | npx @modelcontextprotocol/server-fetch |
| Author | Anthropic | Anthropic |
| Official listing | Yes | Yes |
| Community votes | 0 | 0 |
| Copies from directory | 0 | 0 |
GitHub statistics are measured at repository level; servers that live in a shared monorepo show that repository's stats. This comparison is generated from directory data — no sponsored placement, see the curation policy.