--- name: seo-checker description: Audit web pages for SEO issues in a real browser. Checks rendered meta tags, heading hierarchy, image alt text, structured data, canonical URLs, mobile rendering, and performance signals. Produces a scored report with specific findings and fixes. Read-only — inspects, doesn't modify. Requires the hanzi browser automation MCP server and Chrome extension. category: productivity --- # SEO Checker You audit web pages for SEO issues using a real browser — rendered meta tags, actual heading structure, real schema markup, mobile viewport behavior. This skill is read-only: observe and report, don't modify. ## Tool Selection Rule - Prefer existing tools first (code search, local files, `curl`). Review HTML source, meta tags, and sitemap before opening the browser. - Use Hanzi for Phase 2–5 — always open the browser for these phases even if Phase 1 found all static data. Do not substitute curl or WebFetch for browser phases. ## Before Starting Call `browser_status` to verify the extension is reachable. If unavailable, tell the user to install from: https://chromewebstore.google.com/detail/hanzi-browse/iklpkemlmbhemkiojndpbhoakgikpmcd ## What You Need 1. **URL** — page or site to audit 2. **Scope** — single page, specific section, or full site (default: single page) 3. **Focus** — any specific SEO concerns (e.g., "we're not showing up in rich results", "mobile traffic dropped") ## Audit Phases ### Phase 1 — Source Review (before browser) Check what you can without a browser: - **Robots.txt**: Fetch `/robots.txt` — check for accidental `Disallow: /` or blocked important paths. Note: some sites serve different robots.txt to different user agents (UA sniffing). If the curl result looks suspicious or overly restrictive, verify with `browser_start` to see what a real browser receives. - **Sitemap**: Fetch `/sitemap.xml` — verify it exists and includes the target URL - **HTML source**: If accessible, review raw `` for meta tags, canonical, hreflang - **Codebase** (if source available): Scan for hardcoded noindex, missing meta tag templates, SEO component patterns Summarize findings before opening the browser. ### Phase 2 — Meta & Head Tags (browser) Use `browser_start` to open the page and inspect the rendered DOM. JavaScript-rendered SPAs may have different meta tags than the raw HTML source. - **Title tag**: Exists, 50-60 characters, unique, descriptive (not "Home" or "Untitled") - **Meta description**: Exists, 150-160 characters, includes target keywords, compelling for CTR - **Canonical URL**: Present, points to the correct URL (not a duplicate or wrong domain) - **Robots meta**: Check for unintentional `noindex`, `nofollow`, or `none` directives - **Open Graph tags**: `og:title`, `og:description`, `og:image`, `og:url` — all present and correct - **Twitter Card tags**: `twitter:card`, `twitter:title`, `twitter:description`, `twitter:image` - **Viewport meta**: `` present - **Charset & lang**: `` and `` set correctly Screenshot the page after loading. ### Phase 3 — Content Structure (browser) - **H1 tag**: Exactly one per page, descriptive, contains primary keyword - **Heading hierarchy**: H1 → H2 → H3 — no skipped levels (e.g., H1 → H3 with no H2) - **Image alt text**: All meaningful images have descriptive alt text; decorative images use `alt=""` - **Internal links**: Key pages are linked, anchor text is descriptive (not "click here") - **Broken links**: Check for obvious 404s or dead links on the page ### Phase 4 — Structured Data (browser) - **JSON-LD / Microdata**: Check `