Netlify is a Deployment MCP server that lets Claude Code, Cursor, Windsurf and any MCP-compatible AI agent site deploys, functions, forms. Install in 1 minute with mcpizy install netlify.
mcpizy install netlifynpx -y @netlify/mcp-serverNetlify MCP exposes the Netlify API as agent tools: `list_sites`, `get_site`, `list_deploys`, `get_deploy`, `trigger_build`, `get_function_logs`, `update_site_settings`. Authentication is a Netlify personal access token. The official MCP from the Netlify team covers the dashboard's most common operations and integrates cleanly with the Netlify CLI for local-to-prod flows.
We use Netlify MCP for two main workflows. First, Jamstack site triage: a build fails on a Hugo or 11ty rebuild, the agent fetches build logs, identifies the broken page or partial, and drafts a fix. Second, A/B and split-testing setup: the agent reads the current split-testing config, proposes branch + traffic allocation, and pushes the change via `update_site_settings`. Token cost is lean — a build log for a typical static site is ~1-2k tokens, function logs for an hour run ~2-4k.
Compared to Vercel MCP, Netlify is the right pick when the workload is static-site-heavy (Hugo, 11ty, Astro static export) or when you need Netlify-specific primitives like split testing, form handling, or Identity. Vercel wins for SSR-heavy Next.js workloads. Compared to Cloudflare Pages MCP, Netlify wins on ecosystem maturity (every static-site generator has a Netlify deploy adapter); Cloudflare wins on raw performance and pricing at scale. The honest trade-off: Netlify's free tier has tight build-minute caps (300/month) — an agent that triggers builds aggressively can exhaust them on a single afternoon.
Site IDs are UUIDs, not slugs. `get_site` with the site name fails — the agent has to call `list_sites` first and pin the ID. Cache it in the system prompt for sessions.
`trigger_build` does not pass through environment variables defined in the dashboard if the build is triggered without a branch context. For agent-driven builds, always pass an explicit `branch` parameter.
Function logs (Netlify Functions / Edge Functions) are streamed only — `get_function_logs` returns the last hour at most, and there's no equivalent of Vercel's `since` parameter on older MCP versions. For longer-window debugging, configure a log drain to an external collector.
Rate limits are not publicly documented but the MCP will return 429s if the agent polls `list_deploys` faster than ~1 req/sec sustained. Use webhooks (configured in the dashboard) instead of polling for deploy completion.
Honest pros/cons against the closest deployment MCP servers.
| Server | Strengths | Trade-offs |
|---|---|---|
| Vercel MCP | Better for Next.js / SSR workloads, polished build-log API | Pricier at scale, no built-in split testing |
| Cloudflare Pages MCP | Cheaper at scale, edge-first runtime | Smaller ecosystem for static-site adapters |
| GitHub Pages MCP (community) | Free, zero-config for plain static hosting | No serverless functions, no preview deploys per PR |
If Netlify doesn't fit your stack, these Deployment MCP servers solve similar problems.
The Netlify MCP server is an Deployment Model Context Protocol server that lets Claude Code, Cursor, Windsurf, VS Code with Copilot, and other MCP-compatible AI agents site deploys, functions, forms. It exposes Netlify's capabilities as tools the AI can call directly from your editor or CLI.
The fastest way is the MCPizy CLI: run `mcpizy install netlify` and MCPizy will add the server to your `.claude.json` automatically. You can also install it manually by adding an entry under `mcpServers` in `.claude.json` with the command `npx -y @netlify/mcp-server` and restarting Claude Code.
Yes. The Netlify MCP server is free and open source (see the GitHub repository linked on this page). You may still need a Netlify account or API key to connect the server to the underlying service, but the MCP layer itself has no MCPizy subscription cost.
Yes. Any MCP-compatible client works — including Claude Code, Claude Desktop, Cursor (via `.cursor/mcp.json`), Windsurf, VS Code with Copilot Chat, and custom agents built on the MCP SDK. The same install command targets all of them; only the config file path differs.
Once installed, your AI agent can site deploys, functions, forms directly inside your conversation. Typical use cases include asking Claude Code or Cursor to run Netlify operations, inspect results, chain Netlify with other MCP servers (see our Workflow Recipes), and automate repetitive deployment tasks without leaving your editor.