ax-audit checks website readiness for AI agents

What happened

Lucio Duran released ax-audit, a command-line tool that audits websites on 18 checks (14 weighted, 4 informational) covering LLMs.txt, robots.txt with content signals, HTML rendering, structured data, HTTP headers, Agent Card, MCP, SEO basics, security.txt, meta tags, OpenAPI, TLS/HTTPS, sitemaps, and content negotiation for AI agents. The tool produces a score from 0–100 and links each finding to a remediation guide.

Why it matters

AI agents and LLMs are increasingly crawling and interacting with websites. Just as Lighthouse audits web performance and axe-core audits accessibility, ax-audit tells site operators how ready they are for the AI agent ecosystem—revealing failure modes invisible to normal website operators, like a Web Application Firewall blocking crawlers that robots.txt actually allows.

What to watch

The tool runs via command line (npx ax-audit https://example.com), supports batch audits with parallel execution, outputs to terminal, Markdown, JSON, or HTML, and gates CI with exit codes (0 for score ≥70, 1 below). Documentation and remediation guides are available at lucioduran.com/projects/ax-audit/docs. The project is open-source under Apache 2.0 and accepts contributions.