Umbraco has launched Backoffice Extension Skills for its open-source content management system, aimed at developers using AI agents to customise Umbraco projects.
The new tools are based on Anthropic's Agent Skills standard, introduced in late 2025 and later published as an open standard under the Linux Foundation. They are designed to direct AI systems to the latest software documentation, examples and source code when developers use assistants such as Claude, ChatGPT or custom LangChain agents.
That is particularly relevant for a platform whose users often tailor websites and digital services through custom dashboards, property editors and other backoffice changes. All of Umbraco's backoffice extension points are already registered, documented and supported, and the new skills are intended to mirror that structure for AI-driven development work.
In practice, the skills are exposed through a standardised endpoint that AI tools can call when completing tasks in Umbraco projects. This is intended to reduce the risk of agents relying on outdated material or older frameworks when producing code or implementation suggestions.
Phil Whittaker, Staff Engineer, AI at Umbraco, described the launch as a response to changes in how developers are starting projects and testing ideas.
"We're seeing a fundamental shift in how digital projects are delivered. AI is becoming the starting point for new projects and is where developers explore ideas. Backoffice Extension Skills ensure that when your team uses an AI assistant, the correct Umbraco knowledge is available," Whittaker said.
Agent shift
The launch reflects a broader move by software vendors to package technical information in formats AI tools can use more reliably. As coding assistants become more common in web development, suppliers have faced pressure to make product documentation easier for AI systems to access and interpret without losing version control or context.
Anthropic's Agent Skills framework is one example of that effort. Since its introduction, the standard has seen broad adoption across Claude and other AI platforms, giving software companies a way to expose task-specific guidance to AI agents through a common interface.
For Umbraco, the focus is on the backoffice, where developers and digital teams often make the most complex custom changes to meet business requirements. Each backoffice extension point can now link AI skills to current documentation, examples and source code, giving agents a more direct path to approved implementation guidance.
Filip Bech-Larsen, Umbraco's chief technology officer, said the model follows the architecture already used inside the platform.
"Agent Skills mirror the way that Umbraco's backoffice extension points work. Each backoffice extension point links AI Skills to the latest Umbraco documentation, source code, and examples. This supports safer development and positions Umbraco as one of the first major CMS platforms to deliver a comprehensive, agent-ready architecture," Bech-Larsen said.
He added: "Umbraco Skills for extending the backoffice is the first series of skills, built to support developers using AI agents for building and extending their digital projects."
Developer use
Some digital agency partners have already been testing AI-based workflows around Umbraco. One example cited is the creation of automatic alt text generators intended to improve website accessibility and search visibility.
Those early experiments point to the kinds of narrow, repeatable tasks where developers and agencies are trying to use AI in production work. In content management systems, that can range from interface extensions and workflow tools to support for editors handling large volumes of digital content.
Umbraco, founded in 2003, is a Microsoft .NET-based open-source CMS with a community of more than 250,000 developers and users, according to the company. It is based in Odense and has offices in the US, UK and Australia, with more than 150 employees.
The release also adds to competition among CMS providers seeking to show their products can work effectively with AI-assisted development. For open-source platforms in particular, one challenge is preserving flexibility for developers while ensuring AI tools rely on maintained documentation rather than fragmented community advice scraped from across the web.
By linking agent workflows to defined extension points, Umbraco is trying to make that process more structured.