
Umbraco CMS MCP Dev
io.github.umbracoPublished by GitHub account umbracov18.0.1Updated Oct 24, 2026
A developer focused model context protocol (MCP) server for Umbraco CMS
Overview
A local MCP server that connects an AI assistant to an Umbraco CMS back office through the Umbraco Management API, using an API user's credentials.
- What you can do
- Manage Umbraco content and back office data through the Management API, including documents, media, document types and data types, by natural conversation instead of the UI. Limit which tool collections are exposed by listing them in a configuration variable. Run in read-only mode so only read and query tools remain available. Check the connected site's version on startup and warn when the package major does not match.
- When to use it
- Useful for Umbraco developers and editors who want an assistant to handle repetitive or complex back office tasks, such as content and media management, without clicking through the UI.
- Requirements
- Runs locally via npx from the npm package, typically configured in Claude Desktop. Needs a reachable Umbraco instance URL in UMBRACO_BASE_URL and an Umbraco API user's client ID and secret in UMBRACO_CLIENT_ID and UMBRACO_CLIENT_SECRET. Optionally set UMBRACO_INCLUDE_TOOL_COLLECTIONS and UMBRACO_READONLY. Pin the package to the site's Umbraco major version.
Installation
In SourceWeft
- Open Umbraco CMS MCP Dev in the dashboard and add it to a workspace.
- Enable the server for the chats that should use its tools.
Desktop only via STDIO. STDIO servers start a local process, so they need the SourceWeft desktop host.
Other MCP clients
Follow the launch instructions in the repository.
README
Umbraco MCP [GitHub License]
An MCP (Model Context Protocol) server for Umbraco CMS that unlocks AI-powered content management. It provides comprehensive access to the Umbraco Management API, enabling your AI agent to perform back office tasks through natural conversation - freeing you from the UI and making complex, repetitive workflows that would be tedious or impossible manually become effortless.
Intro
The MCP server authenticates using an Umbraco API user, ensuring secure, permission-based access to the Umbraco Management API. This means you maintain complete control over what your AI agent can do through Umbraco's standard user permission system - no special security model required.
Quick Start
1. Create an Umbraco API User
First, create an Umbraco API user with appropriate permissions. You can find instructions in Umbraco's documentation.
2. Install in Claude Desktop
Download and install the Claude Desktop app, then add the MCP server to your configuration:
- Open Claude Desktop Settings > Developer > Edit Config
- Add this configuration to your
claude_desktop_config.json:
- Fully restart Claude Desktop (including in the system tray on Windows)
Tip: Pin
@umbraco-cms/mcp-devto your site's Umbraco major (e.g.@17above, or@18for an Umbraco 18.x site) rather than@latest. Tool schemas track the Management API of the package's own major, so a mismatched major can fail confusingly mid-task once a tool's shape has drifted. The server checks the connected site's version on startup and warns on a mismatch, but pinning avoids it altogether.
Documentation
For complete installation instructions, configuration options, tool listings, and usage examples, see the full documentation:
Hosted Deployment
The Cloudflare Worker entry point (src/worker.ts) supports two deployment modes, selected at request time by the @umbraco-cms/mcp-hosted library:
- Single-tenant (default) — the worker authenticates against the
UMBRACO_BASE_URLconfigured inwrangler.toml. This is the existing behavior and requires no extra configuration. - Multi-tenant Umbraco Cloud — set
UMBRACO_CLOUD_ROUTING_ENABLED = "true"under[vars]inwrangler.toml. Requests to/at/{alias}/...resolve the upstream Umbraco Cloud project for{alias}and issue access tokens scoped to that resource per the MCP resource-indicator spec. Each Umbraco Cloud project served by the worker must register an OpenIddict client with idumbraco-cms-developer-mcp-hosted.
siteRouting is wired unconditionally; the per-request toggle is the env var.
Contributing with AI Tools
This project is optimized for development with AI coding assistants. We provide instruction files for popular AI tools to help maintain consistency with our established patterns and testing standards.
Using rulesync
The project includes rulesync configuration files that can automatically generate instruction files for 19+ AI development tools. Generate configuration files for your preferred AI tools:
Other AI Tools
rulesync supports 19+ AI development tools including GitHub Copilot, Cline, Aider, and more. Check the rulesync repository for the complete list of supported tools.
The instruction files cover:
- MCP development patterns and conventions
- TypeScript implementation guidelines
- Comprehensive testing standards with builders and helpers
- Project-specific context and architecture
- API integration patterns with Umbraco Management API
Source: README.md at commit e108eb4
Tools
0Version history
1- v18.0.1LatestSep 16, 2026
Score
50% of the applicable checks met
How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.
- Reached in the last 30 days25 pts
Keep the endpoint answering the MCP handshake; SourceWeft probes it every 30 days.
- Auth declared10 pts
Declare the credential headers in the registry entry, or keep the endpoint answering the probe so its auth can be classified.
- Tools listed15 pts
Answer tools/list without a credential, or ship a package that starts in a sandbox without one.
- DescriptionRequired15 pts
- Logo10 pts
- Documentation15 pts
- License10 pts