netlify-mcp (Python)

A Python port of the official Netlify MCP server, built to behave identically to the TypeScript original for the local stdio use case: same tool names, same input schemas, same response text.

This port targets the local stdio server only. Two things in the original are intentionally out of scope:

• The remote proxy deploy mode (--proxy-path / JWE flow), used when the MCP server runs behind Netlify's hosted remote endpoint.
• The full remote OAuth/OIDC hosting server.

Everything else is ported: the user, team, project, deploy, and extension domain tool groups, the netlify-coding-rules context tool, and the Claude Design "Send to Netlify" import tools (import-claude-design-from-url, get-design-import-job-status).

Requirements

• Python 3.12+
• A Netlify account, authenticated via one of:
  • netlify login (using the Netlify CLI) — this server reads the same config.json the CLI writes.
  • The NETLIFY_PERSONAL_ACCESS_TOKEN environment variable, which overrides the CLI-managed token when set.

Install and run

uv sync
uv run netlify-mcp

Or, for one tool per operation instead of the grouped read/write selector tools:

uv run netlify-mcp --verbose

MCP client configuration

{
  "mcpServers": {
    "netlify": {
      "command": "uv",
      "args": ["run", "--directory", "/path/to/python-netlify-mcp", "netlify-mcp"]
    }
  }
}

Tools

By default, each domain exposes two selector tools: netlify-<domain>-services-reader (read-only operations) and netlify-<domain>-services-updater (write operations). Each selector takes a selectSchema of {operation, params} covering every operation listed below for that domain. Pass --verbose to instead register one tool per operation, named netlify-<domain>-<operation>.

user

team

project

deploy

extension

Standalone tools

These are registered individually, not as part of a domain's reader/updater selector, and are unaffected by --verbose.

Configuration

Environment variables (see .env.example):

Development

uv sync --group dev
uv run pytest
uv run ruff check .
uv run mypy src