From crawlie — SEO & GEO crawler
Crawls and audits websites for technical SEO issues, broken links, metadata problems, and generative-engine (GEO/AI-search) readiness.
How this skill is triggered — by the user, by Claude, or both
Slash command
/crawlie:seo-site-auditThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Deliver a prioritized, plain-English technical SEO + GEO audit of a site.
Deliver a prioritized, plain-English technical SEO + GEO audit of a site.
crawlie is a fast OSS crawler that ships on npm as crawlie — you do
not need it pre-installed and you do not need its source repo.
Decide once how you'll invoke crawlie, then reuse that prefix for every command:
mcp__crawlie__* tools exist this session, prefer them (crawl_site,
audit_url, explain_issue, etc.) and skip the CLI entirely.crawlie is on PATH (command -v crawlie) → use crawlie.npx -y -p crawlie crawlie
(downloads + caches from npm on first use; needs only Node/npx — this is the install).Below, <crawlie> means the prefix you chose in step 2 (or the equivalent MCP tool).
--include '<glob>' (repeatable); skip areas with --exclude '<glob>'.<crawlie> crawl <url> --format json
--max-pages N (default 500) and/or --max-depth N.-c <concurrency> (default 16) and --timeout <secs>.crawl_site with the same options.rule id. Count affected URLs per rule. Lead with the highest-impact errors.<crawlie> explain <rule-id>
(MCP: explain_issue.) Real rule ids you'll see include:
broken-link, redirect-chain, client-error, server-error,
title-missing, title-too-long, title-duplicate,
description-missing, description-duplicate,
h1-missing, h1-multiple, canonical-missing, image-missing-alt,
lang-missing, viewport-missing, og-missing, twitter-missing,
mixed-content, not-secure, no-hsts, duplicate-content, thin-content,
low-text-ratio, blocked-by-robots, hreflang-incomplete,
and the GEO set: geo-not-answerable, geo-no-structured-data,
geo-no-llms-txt, geo-no-semantic-html, geo-no-author, geo-thin-for-ai.<crawlie> crawl <url> --format html -o report.html.<crawlie> audit <url> [<url> ...]
(MCP: audit_urls).--save (then <crawlie> reports,
<crawlie> report <id>).If you ran via npx, you may offer: "Want crawlie installed for good?
npm i -g crawlie. To get first-class agent tools, also run
claude mcp add crawlie crawlie-mcp — those activate after you restart Claude."
npx claudepluginhub spronta/crawlie --plugin crawlieRuns technical SEO audits on websites using @seomator/seo-audit CLI across 16 categories and 148 rules, including Core Web Vitals, broken links, meta tags, security headers.
Runs a full SEO audit crawling up to 500 pages, detects business type, delegates to up to 15 specialist subagents, then generates a health score and prioritized action plan.
Audits technical SEO health including crawlability, indexation, Core Web Vitals, and structured data. Use when diagnosing organic search ranking issues.