How to Install squirrelscan: A Free Website Audit Tool You Run in One Command
Tools
Step-by-step
Squirrelscan
Trigger Marketing
The Trigger team
⊹
Paid SEO audit tools love a monthly invoice. squirrelscan does the same kind of check for exactly zero dollars, from one command, and then hands you the fixes like a polite little robot butler.
Let's hire the butler.
It crawls a site and checks it against 300+ rules in 22 groups: crawlability, core SEO (titles, H1s, meta descriptions), "Agent Experience" (do AI bots like GPTBot and Claude-User get the same content a browser gets?), structured data, local SEO, E-E-A-T signals, links, speed, security and accessibility. Every problem comes with a fix you can paste straight into a coding assistant.
Quick words (one line each):
CLI (command-line tool): a program you run by typing in a terminal instead of clicking buttons.
npm: the installer that comes with Node.js. It downloads tools for you.
H1: the main headline of a page. One per page.
Telemetry: a tool quietly sending usage info back to its makers.
E-E-A-T: Google's shorthand for experience, expertise, authority and trust.
Step 1: Do a 60-second safety check (no skipping, it's 60 seconds)
Before you install anything from GitHub, look at four things. Here is what squirrelscan showed when we checked it on 2026-10-26:
Licence: MIT. You can use it, change it and even sell work made with it. Keep the copyright notice if you reuse its code. (The name and logo are trademarks, so don't slap them on your own product.)
Last update: 2026-10-26. It's very active, with 100+ commits since late July.
Stars: 299 (as of Oct 28, 2026). Small but growing. Stars mean people use it, not that it's safe.
Install method: the npm package downloads one ready-made program from the project's GitHub releases and checks its SHA-256 hash (a fingerprint that proves the file wasn't swapped). That's the good kind of installer.
Step 2: Install it
You need Node.js (which gives you npm). Then, in a terminal:
npm install -g squirrelscan
The program you run after that is called squirrel.
Just want to try it once? Skip the install:
npx squirrelscan audit example.com
They also offer curl ... | bash and iwr ... | iex one-liners. Those download a script and run it before you've read it. If you use them, open the script first. The npm route above is the one we tested.
Step 3: Turn off telemetry (be a little private)
Telemetry is on by default. It's small and anonymous, but you're here for an audit, not to be counted. Pick one:
squirrel self settings set telemetry false
or set NO_TELEMETRY=1 in your terminal before you run it (in PowerShell: $env:NO_TELEMETRY=1).
Step 4: Run a quick audit on 5 pages
squirrel audit https://yoursite.com --max-pages 5 --level quick -f markdown
--max-pages 5 keeps it small and polite.
--level quick runs the fast check.
-f markdown gives you a readable report.
On our real 5-page quick test it finished in 6.9 seconds, found 19 URLs in the sitemap and checked 5 of them. Quicker than making coffee. Much quicker than reading a 90-page agency PDF.
Only point it at sites you own or have permission to check. It crawls whatever you give it.
Step 5: Read the report (the good stuff first)
The report gives an overall score plus a score per group. In our test the findings were real and useful. It caught:
A home page with no H1.
A broken internal link (a page linking to a URL that returns 404, meaning "not found").
Template leftovers: every checked page still carried the old website template's title and meta description, plus the template's demo email and phone links. That's the kind of thing a human stares past for months. The robot does not blink.
Bloated pages: hidden elements stuffed with links and text, and raw HTML so heavy that visible text was under 15% of it.
If you built your site from a template, run this today. Seriously. Template ghosts are real.
Step 6: Hand the fixes to your AI assistant
Switch the output to the format made for AI assistants:
squirrel audit https://yoursite.com --max-pages 5 --level quick -f llm
Paste that into Claude, Cursor or another coding assistant. Each issue comes with a fix prompt, so your assistant knows exactly what to change.
Want it built in? This adds two skills (squirrelscan and audit-website) for Claude Code, Cursor or Codex:
squirrel skills install
Step 7: Ignore the warnings that don't earn their keep
Not every warning deserves your afternoon.
"No /llms.txt found": skip it. Peec AI tested llms.txt and similar files and saw no visibility gain (15:30-16:30, Sept 30, 2026). Victorious says skip llms.txt by default (31:12-32:44, Oct 1, 2026).
"noai" and licensing signals: treat as info only. Google Search Central says made-up robots.txt rules like "no-ai" are ignored; only user-agent, allow, disallow and sitemap count (3:02-3:32 and 7:06-8:00, Oct 21, 2026).
AGENTS.md checks: same bucket. Info, not a must-do.
Step 8: Know what's free and what isn't
Local CLI: free. MIT, no account, no API key.
Cloud extras: sign in with squirrel auth login (free or Pro account) for cloud browser rendering, AI page analysis, scheduled audits, shareable reports and link monitoring. Some of those use credits. You don't need any of it to run local audits.
Do this today
Run npm install -g squirrelscan.
Turn telemetry off: squirrel self settings set telemetry false.
Run squirrel audit https://yoursite.com --max-pages 5 --level quick -f markdown.
Fix every missing H1, broken link and leftover template title it finds.
Ignore the llms.txt warning. You have better things to do.
Related reading
Sources
Don't worry if this seems like it's a pain in the ass for someone who doesn't build websites. That's because it is. The good news: just click the link below, pull the trigger, and we'll take care of the rest.
Share This Article
