Docsy Moves to Linux Foundation, Adds AI-Readability Scoring for Technical Docs
Google's documentation platform is joining the Linux Foundation and introducing metrics to measure how well AI agents can parse technical documentation.
Machine learning systems are now consuming technical documentation at scale, forcing projects to rethink how they structure and publish information. Docsy, an open source documentation framework created by Google and widely adopted across cloud native projects, is transitioning to the Linux Foundation while rolling out new capabilities designed specifically for AI readers.
Erin McKean, senior developer relations engineer at Google and member of the Docsy steering committee, unveiled the transition during a keynote presentation Wednesday at the Linux Foundation's Open Source Summit Europe in Prague.
Since its introduction in 2019, Docsy has served as an open source theme for the Hugo static site generator, tailored for technical documentation. While applicable to any documentation need, including proprietary systems, the project has found its strongest foothold in the open source community. By the close of 2024, approximately 2,200 projects had adopted Docsy, with significant usage among Cloud Native Computing Foundation members such as Kubernetes, OpenTelemetry, gRPC and Jaeger.
The relocation to the Linux Foundation reflects the existing concentration of Docsy users within the foundation's ecosystem. McKean explained that the move brings the project into closer alignment with its user base. "Open source projects work best when they are close to the users," she stated.
Documentation as infrastructure for AI
McKean's keynote emphasized the emergence of artificial intelligence as a significant new audience for technical documentation. While acknowledging that some technical writers feel overlooked by the sudden influx of resources driven by AI adoption, she stressed that the ultimate goal remains getting information into the hands of developers, regardless of the delivery mechanism.
When we're making technical documentation, it really doesn't matter how the information becomes useful to humans, as long as it does it.
Erin McKean
She elaborated with a hypothetical: "If someone told me that there was evidence that said opera is the best way to reach your project users, I'd be writing operas."
Docsy has already begun modifying its output to accommodate machine learning tools. Beginning with version 0.15.0 in May, the platform generates a Markdown rendering of each page in addition to standard HTML output, and produces an llms.txt file that provides AI systems with a searchable index of site content. Both capabilities are optional and currently marked as experimental.
The underlying strategy is to establish a clearer pathway for AI systems to access the specific information projects intend them to use.
You can redirect your LLMs and agents to the text that tells them how to use the project.
Erin McKean
Docsy has continued expanding its machine-readable features. Version 0.16.0, shipped in July, introduced an upgrade guide formatted for AI assistants to follow, incorporating structured conditions, steps and verification checkpoints. Version 0.17.0, released in August, advanced agent compatibility further by automatically inserting a hidden directive at the top of each page on sites with llms.txt enabled, guiding visiting agents to the site's content index. This feature remains experimental.
Measuring agent-friendliness
The roadmap ahead includes the introduction of agent-friendly, or "AF," documentation scores, a metric system intended to help project maintainers evaluate how effectively AI tools can locate, traverse and process their documentation. McKean indicated this would supply teams with concrete targets for improvement.
You won't have to guess. You can measure how agent friendly your docs are.
Erin McKean
Beyond serving AI systems, enhanced documentation carries practical benefits for human maintainers. McKean pointed out that comprehensive documentation can preempt routine support requests, lightening the workload on project teams.
When you have good docs, it reduces the number of questions that can be easily answered by documentation, reducing the burden on maintainers.
Erin McKean