About findable-audit

findable-audit is a free, open-source tool that measures how findable a website is — by classic search engines and by AI assistants such as ChatGPT, Claude and Perplexity.

It answers one question: when someone searches for what you offer, can search engines and AI crawlers reach your pages, extract your content and cite you as a source?

What it checks

Each audit samples several pages of the target site and runs 137 checks grouped into 8 weighted families: AI access (robots directives for GPTBot, ClaudeBot and friends), answer-engine content (llms.txt, extractable server-rendered copy), structured data (JSON-LD entities, Open Graph), technical SEO (canonicals, redirects, sitemaps), on-page semantics (titles, headings, internal links), performance and Core Web Vitals, accessibility, and security headers.

How scoring works

Every check awards points and ships with a concrete, prioritized fix. Family scores are weighted into a single 0–100 score and an A–F grade, so two audits of the same site are directly comparable over time — the CLI can even fail a CI build when a deploy regresses the score.

Open source

The engine is MIT-licensed and dependency-light. The same check catalogue powers this web app, a command-line tool and a CI gate. This site is audited with its own engine — dogfooding keeps the recommendations honest.

The 8 weighted families

AI access (weight 16%): robots.txt directives and edge rules for GPTBot, ClaudeBot, PerplexityBot and 25 other AI agents.
Answer-engine content (18%): llms.txt, server-rendered copy, direct-answer leads, extractable lists and tables.
Structured data (15%): JSON-LD entities, Open Graph, FAQPage and Article markup, entity grounding through sameAs.
Technical SEO (15%): canonicals, redirects, sitemaps, hreflang, host consolidation and pagination.
On-page (12%): titles, meta descriptions, heading hierarchy, image alt text and internal linking.
Performance and Core Web Vitals (10%): LCP, CLS, INP and TTFB, measured through PageSpeed Insights when an API key is supplied.
Accessibility (7%): landmarks, form labels, iframe titles, descriptive alt text and heading order.
Security and trust (7%): HTTPS, HSTS, Content-Security-Policy, X-Content-Type-Options and Referrer-Policy.

Frequently asked questions

Is findable-audit free?

Yes. The web app and the command-line tool are both free and MIT-licensed, and no account is needed to run an audit.

How is GEO different from SEO?

SEO optimizes for ranked links on a results page. GEO optimizes for being reached, extracted and cited inside an AI answer. findable-audit scores both in one pass, because a single crawl answers both questions.

What do you keep about the sites I audit?

The report itself is ephemeral and expires from the server a few minutes after the audit ends. One minimal usage record per audit is kept: the URL and its domain, the score, the grade, the eight family sub-scores, a timestamp and a salted hash of the IP address. Page contents and findings are never stored.

Can I run findable-audit in continuous integration?

Yes. The CLI stores a baseline report and, with --fail-on-regression, exits non-zero when a deploy lowers the score. A composite GitHub Action and JUnit XML output ship with the repository.