a14y

a14y is an open source project focused on improving the readability of websites for AI agents. It addresses the challenge that AI agents now read and process much of the web before humans do, and provides a structured way for site owners and developers to measure and enhance how well their sites can be discovered, parsed, and comprehended by these agents. The platform is built around an open specification and a versioned scorecard, which together define a set of 38 checks that assess agent readability.

Key features include a command-line interface (CLI) tool, a Chrome extension, and a public leaderboard. The CLI allows users to audit any page or entire site from the terminal, producing scored reports in text, JSON, or markdown formats suitable for further automation or review. The Chrome extension enables in-browser scoring of any page, with the ability to switch between scorecard versions and access direct guidance for fixing failed checks. Additionally, a14y offers a spec-compliant agent skill that can automate the audit and remediation process, compatible with coding agents that support the open skills format.

Each check in the scorecard is thoroughly documented, providing details on what is being evaluated, why it matters for agent comprehension, and step-by-step instructions for remediation. txt file, which serves as an agent-friendly equivalent of a sitemap, guiding AI crawlers to the most relevant pages. The tools can be run in various environments, including CI pipelines and headless setups, and are designed to be fast and developer-friendly.

0 license and was created by Timothy Jordan. All tools and specifications are openly available, and the project maintains transparency through its public leaderboard and open documentation. The platform is suitable for web developers, site owners, and anyone interested in optimizing their web content for AI agent accessibility and comprehension.