NetBramha Studios · Internal Tool
Site Intelligence Tool

SiteScope

One URL in. Page count, template count, and a first-pass information architecture out — before the pitch, in under four minutes.

< 4
Minutes per audit
3
Discovery sources
0
Manual steps
AI
Template classifier
How it works

Three sources. One merged result. No manual crawling.

01
Sitemap check
Reads sitemap.xml and sitemap_index.xml. If found — instant full page list, no crawling needed.
⚡ instant
→
02
Common Crawl
3 monthly indexes queried in parallel. Bypasses bot protection — fetches CC, not the site.
🌐 no blocks
→
03
Live validation
Bulk HEAD-checks every URL. Dead pages stripped. Only live URLs reach the template classifier.
✓ accurate
→
04
Claude classifies
Full URL list + page signals sent to Claude. Named, described, complexity-rated templates returned.
🧠 AI-powered
The page count question always comes up in a pitch. The template count — which is what drives design effort — almost never gets answered until the SOW is signed.

A redesign pitch has two numbers that matter: how many pages exist (content migration effort) and how many distinct layouts those pages use (design effort). They're rarely the same number. A 300-page site might run on 11 templates.

Before SiteScope, getting those numbers meant two hours of manual crawling, a spreadsheet, and a rough count that varied by who did it last. Now it's a URL and a button.

The template classification is the part that's hard to do manually. Claude receives every URL on the site alongside structural signals from representative pages — headings, forms, tables, downloads — and groups them by visual and functional similarity, not by URL pattern. The result matches what a senior UX designer would produce after an hour of manual review.

What you get

Four outputs from one analysis.

📐
Template classification
Named templates with layout descriptions, anatomy breakdowns, and complexity ratings. /about-us, /csr, /sustainability become one "Editorial" template — not three singletons. Expandable cards, export to Word doc.
🗺
Information architecture
Card-grid IA with section names, page labels, and template tags pulled from the Claude classification. Export opens a print-ready document — a v1 IA for redesign kickoff, not a path list.
📊
Scope summary
Three numbers that mean something: total pages (migration), page sections (navigation architecture), design templates (design effort). The brief translates those into project size in plain language.
🔗
External link map
Cross-domain links grouped by destination. Identifies third-party portals, loyalty microsites, embedded tools — integration dependencies that always surface mid-project if they're not scoped upfront.
The numbers that go into the proposal

Three counts. Three different conversations.

153
Total pages
Every live HTML URL on the domain. Drives content migration estimation — how much content needs to be reviewed, rewritten, or retired.
84
Page sections
Unique first-path segments — the site's navigation architecture. Answers "how many parts does this site have?" before discovery begins.
11
Design templates
Distinct layouts identified by Claude. This is the number that goes into the design effort estimate. Not URLs — templates. The one number the pitch always needs.
Stack

Built the NB way.

Single-file HTML Cloudflare Worker Claude Sonnet Netlify Common Crawl API DOMParser Bulk HEAD validation No framework No build step No login required

Anthropic API key lives in the Cloudflare Worker. Never touches the browser. Worker handles sitemap discovery, Common Crawl queries, bulk URL validation, and Claude API proxying — all in one deployed script.

Built for
Pre-sales site audits
Redesign scope estimation
Pitch preparation
IA discovery kickoffs
Template-level effort scoping