Fetch (reference server)
Browser and web automationFetches a web page and turns it into Markdown for the model
- Stars
- ★ 91.4k
- Licence
- Apache-2.0 / MIT (MCP licence transition)
- Updated
- MCP server
- Official
Microsoft's browser automation server built on Playwright
From GitHub, fetched 19 Oct 2026.
Drives a real browser using the page's accessibility tree rather than screenshots, which makes it fast and predictable. Agents can navigate, click, fill forms and take snapshots.
| Runs as | Local |
|---|---|
| Type | MCP server |
| Maintained by | Official |
| Install | npm: @playwright/mcp |
Sources: registry.modelcontextprotocol.io/v0/servers/io.github.microsoft%2Fplaywright-mcp/versions/latest · github.com/microsoft/playwright-mcp, checked 19 Oct 2026. “Runs as” is from this listing. Blank means we haven't found it in official docs yet, not that it's unsupported.
Fetches a web page and turns it into Markdown for the model
Let a coding agent inspect and debug a live Chrome
Use thousands of ready-made Apify scrapers from an agent
Scrape, crawl and search the web as clean Markdown
Community Playwright server with test-code generation
Search and scrape sites that block ordinary bots