JSON-LD Tester – Structured Data & Agent Tools Inspector

Inspect and test JSON-LD, Schema.org data and agent tools on any website.

Inspect, validate and test structured data and agent tools on any website — built for technical SEOs, web developers, agencies, and teams running shops and SaaS products. New in 2.4 • NEW: dataLayer tab — inspect and validate the page's GA4 events, Consent Mode and GTM containers without the console. • NEW: GA4 event validation — missing required parameters (e.g. purchase without transaction_id), non-numeric value ("129,00"), value 0, currency that is not an ISO 4217 code, items[] without item_id/item_name, events fired twice, legacy Universal Analytics structure. • NEW: Consent Mode v2 — reports missing ad_user_data/ad_personalization, update without or before default, and invalid values. • NEW: PII warning — e-mails, phone numbers, IBANs and card numbers in the dataLayer are detected and reported by path, never by value; they are masked in copies, reports and snapshots. • NEW: cross-check against the JSON-LD on the same page — price, currency and SKU mismatches between items[] and the Product markup. • NEW: snapshots — store locally (masked), filter and compare: added, removed, duplicated and changed events, also against the current page. • Read-only: dataLayer.push is never wrapped, nothing about the page changes. No new permissions. New in 2.3 • NEW: share preview per network – Facebook, LinkedIn, X, WhatsApp, Slack – each with that platform's truncation limits, image ratio and card layout. • NEW: complete Open Graph / Twitter tag checklist (essential and recommended). Twitter tags covered by Open Graph are shown as fallback instead of missing. • NEW: catches the real mistakes – og:url that does not match the canonical, relative or http:// og:image, name="og:…" instead of property= (Facebook ignores it), unknown twitter:card values. • NEW: image check on the measured file (only after you click "load preview") – below 1200×630, below the 200px minimum, odd aspect ratio, declared og:image:width/height that does not match the file. • NEW: hreflang validation – BCP-47 syntax (de_AT, en-UK, a country code used as a language), missing self-reference, missing x-default, duplicates and conflicting targets, relative URLs. • Tools tab: LinkedIn Post Inspector added, the retired X Card Validator link removed. New in 2.2 • New UI: wider window with a left navigation rail. • NEW: On-Page SEO tab – meta tags, headings, Open Graph & Twitter, hreflang, images (alt/title), links (internal/external, unique, nofollow), word count – with actionable issues. • NEW: Headers tab – HTTP response headers, indexability (index/follow), a robots.txt check for the current URL (Googlebot) and HTTP status. • NEW: Tools tab – quick access to common external validation and SEO tools. • NEW: Inspection report as PDF – all sections in a print-ready report (local, nothing uploaded). • Extension name & store listing localized (EN/DE/ES/IT/ZH); side panel auto-updates on tab/domain changes. • No new permissions – every new check is same-origin only. New in 2.1 • NEW: On-Page SEO tab – meta tags, heading outline, Open Graph & Twitter, hreflang, images/alt and links at a glance, with actionable issues. • NEW: Inspection report as PDF – export all sections (Structured Data, On-Page, Agent Tools, Readiness) as a print-ready report. Generated locally, nothing is uploaded. • Extension name & store listing are now localized (EN/DE/ES/IT/ZH). • Fix: the side panel now auto-updates on tab and domain changes. • No new permissions. **Structured Data** - Detect all JSON-LD blocks, including `@graph`, arrays and nested types - Clear valid / warning / error status per block - Tree view and raw view with syntax highlighting and search - Precise parse errors with line and column - Copy JSON, copy the full `<script>` tag, or download - One-click links to the matching schema.org docs - Canonical and robots at a glance - Open Google Rich Results Test or the Schema.org Validator with correct encoding — umlauts, accents, emojis and query strings are preserved **Agent Tools (experimental, WebMCP)** - Detect agent tools a page exposes via `document.modelContext` - See each tool's description, input/output schema, and annotations - Automatic static analysis: naming, descriptions, schema issues, safety - Risk classification: Low / Medium / High / Critical - A safe manual tester that builds a form from the input schema and never runs destructive tools without explicit, multi-step confirmation **Agent Readiness Score** - A local, transparent 0–100 score across structured data, tool availability, descriptions, schemas, safety, testability and semantics - Prioritized High / Medium / Low impact recommendations - Checks agent & crawler conventions: llms.txt, robots.txt AI-crawler access, sitemap discovery, and the ai.txt AI-training opt-out **Privacy first** - No visited URLs, page content or tool schemas are sent to any server - No ads, no tracking, no page modification - Minimal permissions: activeTab, scripting, storage ## Feature list - JSON-LD detection incl. `@graph`, arrays, nested types - Valid / Warning / Error / Experimental / Not-supported status badges - Tree & raw views, search, copy, copy `<script>`, download - Parse errors with line/column - Schema.org doc links; canonical & robots - Correct UTF-8 encoding for Google Rich Results / Schema.org Validator - WebMCP agent-tool detection (experimental) with feature detection - Static tool analysis + Low/Medium/High/Critical risk classification - Safe manual tool tester with risk-gated confirmations - Local Agent Readiness Score with prioritized recommendations - Light/Dark, de/en/es/it/zh, keyboard accessible ## Privacy text This extension does not collect, transmit, or sell any personal data. It analyzes the current page locally when you open it. Visited URLs, page content, and agent tool schemas are never sent to any server. Settings and (optional) test history are stored only in your browser. External validators open only when you click them. ## FAQ **Does it send my data anywhere?** No. Analysis is local; external validators open only on your click. **Why does it need "read the tab"?** Only via `activeTab`, only when you open the popup, to read the page's JSON-LD and detect agent tools. It has no standing access to your sites. **What is WebMCP?** An experimental Chrome API (`document.modelContext`) that lets sites expose tools to AI agents. Support depends on your Chrome version and whether the site opts in; the extension detects it and degrades gracefully. **Will it run a tool automatically?** Never. Tools require explicit confirmation; destructive/critical tools are blocked by default.