An AI Website Audit Tool That Runs the Whole Audit From Your Terminal

If Auditing Your Site Feels Like a Part-Time Job, This One's for You

Let's be honest: nobody opens a spreadsheet at 9pm excited to check meta descriptions. But that's what a real website audit used to look like. You'd crawl with one tool, check broken links with another, eyeball headings manually, and then pay a consultant to tell you what it all means. It's slow, it's expensive, and half the results don't even tell you what to fix first.

That's exactly the pain this project was built to kill. I'm not going to pretend there aren't good paid tools out there — there are. But there's a huge gap between "free but gutless" and "powerful but $200/month," and an automated AI website audit tool should never force you to pick a side.

What Is the AI Website Audit CLI?

The AI Website Audit CLI is an open-source Python tool built by Tools2U that turns any public URL into a structured audit package. Point it at a site and you get four things:

1. Extraction JSON — every fact it found on the page: titles, meta descriptions, headings, links, images, schema types, tech hints like WordPress or Next.js.
2. Deterministic audit JSON — transparent, rule-based checks and scores across seven areas. No vibes, no hallucinated numbers.
3. A Markdown (or HTML) report — a prioritized improvement plan, generated locally or with AI.
4. Optional AI strategy report — powered by OpenAI's Responses API, written strictly from the saved evidence so it doesn't invent problems.

The design philosophy is deliberately boring, and I mean that as a compliment: facts first, rules second, AI last. The extraction and scoring run 100% on your machine — nothing leaves it unless you explicitly ask for the AI report. That's a privacy story most commercial audit SaaS can't match.

Key Features, With Receipts

Anyone can slap "AI-powered" on a landing page. Here's what this automated AI website audit tool actually does, with concrete examples:

Deterministic scoring you can reproduce

Run the audit twice and you get the same scores. That matters more than it sounds, because it means you can track your site's progress over time and trust the delta. A real run on a typical marketing site looks like this:

Each score comes with rule-based findings, severity levels, quick wins, and risk flags — so an 84/100 tells you why, not just what.

Deep extraction, not just a homepage glance

It grabs H1/H2/H3 structure, internal vs external links, broken/empty/JavaScript links, missing alt text, lazy-loaded images without dimensions, CTA detection, Open Graph and Twitter card tags, JSON-LD schema types, word count, text-to-HTML ratio, and privacy signals like cookie banners and pricing pages. There's even an optional same-domain crawl for multi-page checks like duplicate titles and meta descriptions.

An AI report that can't just make stuff up

Here's the part I genuinely like: the AI layer gets the extracted facts and deterministic scores, plus strict system instructions to avoid hallucinated claims. You get a developer implementation checklist and prioritized actions — e.g. "fix the 14 images missing alt text on /pricing before touching anything else" — instead of generic advice ChatGPT would give you about any website ever.

Outputs for every workflow

Markdown for your docs, a standalone HTML report you can send to clients, and JSON for your CI pipeline. If you want to build your own checks, the MIT license means you can fork it and add custom rules. Try doing that with a closed SaaS.

How It Compares to Screaming Frog and Sitebulb

I've used both of the big names, and honestly? They're good. Screaming Frog's crawl depth is legendary, and Sitebulb's visualizations are the prettiest in the business. But here's the thing neither of them does:

If you want the deep-dive on exactly where the CLI wins and loses against these two, the full 2026 SEO audit tool showdown on Notion walks through every category. Spoiler: the CLI wins on price, transparency, and AI, and loses on pure crawl scale — for now.

Quick Start: Your First Automated SEO Audit in 3 Minutes

No dashboard, no onboarding flow, no credit card. You need Python and a terminal. That's it.

pip install ai-website-audit

# Run a fully local audit (no API key needed)
ai-website-audit inspect https://yoursite.com --html

# Add the AI strategy report (needs an OpenAI key in .env)
ai-website-audit audit https://yoursite.com --language en --html

Your output lands in reports/ — an HTML report you can open in a browser, Markdown for your notes, and JSON if you want to automate the whole thing. The inspect command is free and local; the AI report is optional and uses OpenAI's Responses API with gpt-5.5 by default. Set OPENAI_STORE_RESPONSES=false (the default) and even your AI run stays privacy-first.

For a quick reference of every flag and option, there's a community CLI cheat-sheet gist, and if you'd rather see results before installing anything, the interactive demo on Vercel shows sample audit output live.

Who Actually Gets Value From This Tool

• Indie hackers and solo founders who need a real audit before launch without burning a SaaS trial.
• Freelancers and agencies running audits for multiple clients — it's MIT-licensed, so you can rebrand the HTML reports and charge for your analysis, not the tool.
• Developers who want audits in CI. It runs headless, outputs JSON, and plays nicely with cron jobs. Automate a weekly audit and diff the scores.
• SEO folks tired of "AI" tools that are just chat wrappers. If you want evidence you can inspect, this is the closest thing to an open-source SEO audit that competes with paid software.

And if you want to understand the philosophy and roadmap in more depth, the launch write-up on Dev.to covers how the whole thing is built and where it's headed.

Ready to see what an AI website audit tool should feel like? Stop tab-hopping between five free tools and one command-line audit.

Get the AI Website Audit CLI