Google has detailed how it builds, tests, and governs its open-source Agent Skills repository, which has drawn more than 15,000 stars on GitHub. The repository packages Google Cloud domain knowledge into structured instructions for AI coding agents, aiming to make those agents more accurate and reduce errors in development tasks. The project began as a rapid internal effort ahead of Google Cloud Next 2026, with a cross-functional group of Developer Advocates and Technical Writers creating the initial framework. Interest soon spread beyond the original launch group, with engineering and product teams across Google wanting to add skills for their own services. To manage the governance problem created by growth, Google standardised the repository structure for each skill and introduced automated checks before any contribution can be accepted, covering metadata validation, directory layout, naming rules, and line counts. It also tests every URL to catch broken or invented links before a change is merged. Each new skill must go through internal evaluation before public release, with authors providing prompt suites and scoring rubrics. Google runs scheduled weekly checks across the full skills library to detect regressions, comparing agent performance with and without a given skill across different agent frameworks, measuring accuracy and efficiency. Google develops and evaluates skills internally before exporting selected work to the public GitHub repository, with automated export rules stripping out internal assets. It has also created a separate internal programme called DevRel Skills, focusing on agent skills for internal workflows such as content transformation, search engine optimisation and reporting.