Support

squirrelscan ships as a single binary with zero dependencies, so installing is one line.

macOS or Linux:

curl -fsSL https://install.squirrelscan.com | bash

Windows (PowerShell):

iwr -useb https://install.squirrelscan.com/install.ps1 | iex

This drops the binary at ~/.local/bin/squirrel and adds it to your PATH. Run squirrel self doctor afterwards to confirm everything is wired up. Full details live in the installation docs.

Point squirrel at any URL and let it rip:

squirrel audit example.com

squirrelscan crawls the site, runs its audit rules, and prints a health score with the exact issues to fix, and how to fix them. Local audits are free and unlimited, and you don't need an account. Want the report piped straight into your coding agent? Add --format llm:

squirrel audit example.com --format llm | claude

squirrelscan runs 249+ audit rules covering:

• SEO: titles, meta descriptions, canonicals, headings, internal links, and structured data
• Performance: Core Web Vitals, render-blocking resources, and page weight
• Security: HTTPS, security headers, mixed content, and leaked secrets
• Accessibility: alt text, labels, contrast, and landmarks
• Agent experience (AX): how easily AI agents can read and act on your pages

Every run produces a health score, issues ranked by impact, and specific fixes. Browse the full rule list in the rules docs.

Agent experience (AX) is how easily an AI agent, whether a chatbot, a coding assistant, or an AI search crawler, can read, understand, and act on your website. As more of your traffic comes from agents rather than humans, the things that make a page legible to a machine start to matter as much as the things that make it rank.

squirrelscan checks the AX signals that count: clean, parseable HTML, a useful llms.txt, structured data, and content that isn't hidden behind client-side rendering. Curious about the bigger picture? The Nutshell has deeper explainers.

Keeping squirrelscan current is one command:

squirrel self update

That checks for the latest release and applies it in place. Run squirrel self version to see what you're on. Changed your mind? Uninstalling is just as tidy:

squirrel self uninstall

That unlinks the binary and cleans up its cached releases. See the self commands docs for the details.

squirrelscan is built for coding agents like Claude Code and Cursor, and there are three front doors:

• Skills (the fix loop): npx skills add squirrelscan/skills, then prompt your agent to audit a site and fix the issues.
• Local MCP server: claude mcp add squirrelscan -- squirrel mcp runs the audit engine over stdio, no account needed for local audits.
• Hosted MCP server: point any MCP client at https://mcp.squirrelscan.com/mcp, nothing to install.

Your agent then calls audits, rules, issues, and reports as native tools. Full setup is in the agents guide.

Absolutely. The squirrel CLI runs on your machine, so it can audit anything you can reach, including local dev servers:

squirrel audit http://localhost:3000

Auditing a staging site behind basic auth, a preview password, or a bot wall? Pass custom headers with the repeatable -H flag:

squirrel audit https://staging.example.com -H "Authorization: Bearer <token>"

Local audits stay local and free. See auditing projects for more.

Local audits are free and unlimited, forever. The cloud only does what your machine can't: browser rendering for JavaScript-heavy sites, AI analysis, technology detection, and shareable published reports.

Cloud work is priced in credits, and the pricing is flat and predictable:

• 50 credits per audit (covers AI analysis, tech detection, the editor's summary, and publishing)
• 2 credits per rendered page, only when browser rendering is on

Every account gets 500 free credits every month, no card required, and if an audit fails it's refunded automatically. Check your balance any time with squirrel credits. Full breakdown in the credits docs.

Nothing breaks. squirrelscan is designed so a low balance never fails an audit:

• Cloud rules report skipped with a hint, and the audit finishes with the local rules.
• Browser rendering falls back to plain HTTP for the rest of the crawl.
• You get a single one-line warning, not an error.

Local audits never cost credits, so they always run in full. Your free monthly credits reset at the start of each calendar month, and paid plans reset each billing period. Check where you stand with squirrel credits.

The squirrel CLI and all local audits are free, forever. Cloud features run on credits, and the plans are:

Every plan runs the same audit at the same price. Pro adds scheduled audits, custom request headers, higher render concurrency, and credit top-ups; Team adds pooled credits, teammate invites, and roles. Compare them on the pricing page.

Teammates live on the Team plan, which shares one org, one pooled credit balance, and a set of roles (owner, admin, editor, viewer, billing). To set it up, sign in and head to Settings → Team in the dashboard, pick a seat count, and send invites.

A few billing notes: sending an invite costs nothing, seats are billed only once an invite is accepted (2-seat minimum), and mid-cycle seat changes are prorated. See the Team plan docs for roles and rules.

Signed-in audits publish automatically to reports.squirrelscan.com so you can share a link. Reports are unlisted by default: anyone with the link can view, but they aren't indexed. Change that per run with --visibility:

squirrel audit example.com --visibility public

Use public, unlisted, or private, or skip publishing entirely with --no-publish. You can also manage and delete published reports from the dashboard. More in the report command docs.

If an audit comes back with 0 pages, or bails at the very first request, the target site almost certainly blocked the crawler. The usual culprit is a WAF or bot protection returning 403 or 429 at the front door before squirrelscan can read a single page.

A few things to try:

• Pass whatever the site expects with the repeatable -H flag, for example -H "Authorization: Bearer <token>" or a header your firewall allowlists.
• For sites that verify signed crawlers, use Web Bot Auth.
• Confirm the URL loads for you in a normal browser and isn't behind a login wall.

Sites you don't control may simply refuse automated traffic, and that's their call, not a squirrelscan bug.

Pick whichever fits:

• Support ticket (signed in): open one from the support panel in the dashboard and track replies in-app and by email.
• Quick feedback: use the feedback form or run squirrel feedback straight from the CLI.
• Public bugs and feature requests: open an issue on GitHub.

We read everything. The more detail you give us (URLs, report links, what you expected to happen), the faster we can help.