Docsy, the open source documentation theme created by Google, is transferring to the Linux Foundation as it rolls out new capabilities designed specifically for AI agents reading technical documentation, Erin McKean, senior developer relations engineer at Google and a member of the Docsy steering committee, announced Wednesday at the Open Source Summit Europe in Prague. The Hugo static site generator theme, first unveiled by Google in 2019, has become widely adopted for technical documentation across open source projects. By late 2024, roughly 2,200 projects were using Docsy, with major Cloud Native Computing Foundation adopters including Kubernetes, OpenTelemetry, gRPC, and Jaeger.

The platform has begun rolling out features aimed at making documentation more accessible to AI systems. Starting with version 0.15.0 in May, Docsy can produce a Markdown version of each page alongside standard HTML, plus an llms.txt file that provides AI tools with a content index—both features remain opt-in and experimental. Version 0.16.0, released in July, introduced an upgrade guide structured so an AI assistant could follow it, with conditions, steps, and checks embedded in the instructions. The August release of version 0.17.0 went further, automatically adding a hidden directive at the top of each page that directs visiting agents to the site's llms.txt index when that feature is enabled.

According to McKean, the shift to the Linux Foundation positions Docsy closer to the communities already using it, since "open source projects work best when they are close to the users." She emphasized that the rise of AI as a consumer of technical documentation matters primarily for whether information ultimately helps developers, regardless of delivery method. "When we're making technical documentation, it really doesn't matter how the information becomes useful to humans, as long as it does it," McKean said. The ability to point AI systems toward correct information matters because "you can redirect your LLMs and agents to the text that tells them how to use the project," she noted.

The project's roadmap includes "AF," or agent-friendly, documentation scores intended to help maintainers measure how easily AI tools can locate, navigate, and consume their documentation, according to McKean. That should offer projects a concrete benchmark rather than guesswork: "You can measure how agent friendly your docs are." Better documentation also delivers a traditional benefit by reducing routine questions that land on maintainers, since quality docs can answer those questions before manual intervention becomes necessary. Some technical writers have been "a little bit salty" that AI brought more resources to documentation, McKean acknowledged, but the core question remains whether the information reaches developers who need it. For organizations looking to balance automation with developer experience, the underlying tension centers on whether optimizing for machine readers might inadvertently reshape what gets documented and how human audiences ultimately navigate that knowledge.