Model Context Protocol

Give your coding AI
Schema.org Superpowers

Connect Claude Desktop, Cursor, Antigravity, or any MCP-enabled IDE to the JSON Recon engine. One tool call extracts clean, Google-compliant JSON-LD from any publicly accessible URL. $0.01 per extraction.

JSON Recon was built by an SEO practitioner with over 10 years of experience and a deep mastery of schema.org JSON-LD implementations. This isn't a generic scraping API that happens to output structured data. It is a purpose-built structured data engine that specializes exclusively in Google-compliant schema markup: the kind that powers Rich Results, Knowledge Panels, Merchant Listings, and AI-generated overviews. Every extraction rule, every taxonomy decision, and every validation check was designed through the lens of what Google actually consumes, validates, and rewards in organic search.

How JSON Recon Transforms Your Schema Workflow

Structured data is the foundation of modern SEO: Rich Results, FAQ snippets, local map packs, product carousels, and Generative Engine Optimization (GEO) all demand valid JSON-LD. JSON Recon turns that from a manual, error-prone process into an automated conversation with your AI.

🔍

Competitive Schema Audits

Point your AI at a competitor's URL and instantly get their full structured data profile. See exactly which schema types they're using, what properties they populate, and where they're winning Rich Results you're missing. No browser extensions, no copy-pasting. Just ask: "Analyze the schema on this competitor's product page and compare it to ours."

🏪

Local SEO & GBP Alignment

For local businesses, we extract the deepest schema.org subtype. Dentist, not MedicalBusiness. ItalianRestaurant, not FoodEstablishment. Our taxonomy engine selects from 200+ subtypes to ensure your structured data matches Google's entity expectations, improving your visibility in local map packs and AI-powered "near me" queries.

🤖

Generative Engine Optimization (GEO)

AI Overviews and conversational search engines rely on structured signals to rank and cite sources. Pages with comprehensive, valid JSON-LD are significantly more likely to be referenced in AI-generated answers. JSON Recon generates deeply populated schema with sameAs, additionalProperty, and full entity linking, giving AI engines the structured context they need to prefer your content.

Autonomous IDE Workflow

Ask your coding AI: "extract the LocalBusiness schema from this URL and inject it into my header.tsx". It does everything: reads the publicly accessible page, identifies the entity, builds a comprehensive JSON-LD block with address, hours, geo coordinates, and reviews, then writes it directly into your code. The target URL must be live and publicly accessible (not behind authentication or VPN). One sentence. Done.

Save Tokens. Save Compute. Save Hours.

Without JSON Recon, your AI assistant has to build a scraper from scratch, download HTML, parse the DOM, figure out what entity type the page represents, look up the schema.org spec, and assemble the JSON-LD, all inside your context window. Here's what that actually costs:

❌ Without JSON Recon

The Expensive Way

  • 15,000–80,000 tokens of raw HTML dumped into context
  • 3–5 LLM round trips to scrape, detect, extract, validate
  • $0.10–$0.50+ in API costs per page (GPT-4 / Claude)
  • 60–120 seconds of wall-clock time
  • • Risk of hallucinated addresses, invented phone numbers
  • • Often selects generic Organization instead of the correct subtype
✅ With JSON Recon MCP

The Smart Way

  • ~800 tokens of clean JSON returned to context
  • 1 tool call. The MCP server handles everything
  • Only $0.01 per extraction
  • 1–8 seconds total turnaround
  • • Zero hallucination: only real data from the actual page
  • • Always picks the deepest taxonomy subtype (200+ types)
  • • Full provenance metadata: confidence, source, timing

That's a 90% reduction in context window usage and a 10–50x cost saving per page.

Engineered for Production SEO

Schema extraction isn't reliable when your only tool is a single HTTP request. Websites use JavaScript rendering, bot detection, CAPTCHAs, and lazily-loaded JSON-LD. JSON Recon handles all of this automatically, so you don't have to think about it.

🔄

Adaptive Extraction

Our engine automatically handles whatever a page throws at it. Static HTML, JavaScript-rendered SPAs, and heavily protected enterprise sites are all covered seamlessly, with no configuration required from you.

🛡️

Enterprise Site Coverage

Many extraction tools fail on bot-protected sites. JSON Recon successfully extracts structured data from sites that most tools can't reach, giving you reliable results across the web.

🧠

AI-Powered Generation

When no JSON-LD exists on a page, our engine reads the visible content and generates comprehensive schema.org from scratch, selecting the most specific entity type from 200+ options.

Auto-refund on failure: If a website's firewall blocks extraction entirely, your credit is automatically refunded to your account. You never pay for data we couldn't retrieve.

Two-Tool MCP Architecture

JSON Recon gives your AI two tools. The primary tool handles everything intelligently in a single call. The second is a free pre-flight check to assess extraction difficulty before committing credits.

Primary Tool · $0.01

extract_schema

One intelligent tool that handles everything. It first checks for existing JSON-LD on the page. If found and comprehensive, it returns immediately. If the page has no schema or only generic markup, it automatically escalates to AI-powered generation. Built-in advisory engine detects when native schema describes a page section (like an FAQ) rather than the page's primary purpose, and suggests a better type with ready-to-use JSON-LD. Flat rate: $0.01

// Every response includes provenance metadata:
schema_type: "Recipe"
confidence: "high"
source: "json-ld (native)"
extraction_ms: 214
Pre-flight · Free

check_extractability

A free pre-flight check that tells your AI whether a URL is extractable before spending credits. Returns the domain's difficulty rating, expected schema type, known WAF issues, and estimated confidence. Smart agents use this to avoid wasting credits on sites known to block scrapers.

// Pre-flight response example:
extractable: true
difficulty: "easy"
expected_type: "Recipe"
cost: $0.00
$0.01
Per Extraction · Flat Rate

Adaptive pipeline automatically checks for native JSON-LD, escalating to AI-powered entity generation if missing. Same price regardless of complexity.

Specialized in What Google Actually Rewards

Generic AI can write JSON-LD, but it doesn't know what Google validates. JSON Recon does.

🎯

Deepest Taxonomy Selection

We don't stop at LocalBusiness. Our engine maps to 200+ schema.org subtypes: Dentist, AutoRepair, Brewery, NailSalon, and more. Matching exactly what Google's Rich Results tests expect.

🛡️

Anti-Hallucination Guardrails

Our AI is allowed to upgrade taxonomy (choosing the right @type), but strictly forbidden from inventing addresses, phone numbers, prices, or hours. If data isn't on the page, it's not in the schema. Period.

📋

Template Mode for Developers

Missing a required property? Instead of omitting it, the AI inserts instructional placeholders like "REQUIRED: Add product price" so the developer knows exactly what to fill in for Google compliance.

Validated Output

Every response includes proper @context, @type, confidence scoring, and provenance tracking. Output is ready to paste into your <head> or validate in Google's Rich Results Test immediately.

Schema Advisory Engine

Most websites implement schema.org for only one purpose — FAQ markup, breadcrumbs, or a generic organization block. But a single page often serves a deeper purpose that native markup doesn't capture. JSON Recon now detects this gap automatically.

⚠ The Problem

Schema Mismatch Is Invisible

A Laravel development agency's service page has native FAQPage markup — but the page's actual purpose is describing a service offering. The FAQ is just one section. Without advisory, your AI returns the FAQ data and moves on, missing the Service schema that would actually drive rich results for the page's core value proposition.

✅ The Solution

Automatic Type Recommendation

JSON Recon's advisory engine detects when native markup represents a section rather than the page's primary entity. It automatically suggests a better schema type, explains why, and returns the full extracted JSON-LD for the suggested type — ready to implement. One extraction, two schemas.

extract_schema response with advisory
// Native schema returned as-is:
schema_type: "FAQPage"
data: { ...6 FAQ questions... }
// Advisory engine suggests a better type:
suggested_type: "Service"
suggested_reason: "Page primarily describes Laravel
  development consulting services. FAQPage
  only covers the FAQ section."
suggested_data: { ...full Service JSON-LD... }
🔍

Why SEOs Need This

A page with only FAQPage markup misses Rich Result opportunities for its core entity. The advisory engine catches these gaps that even expert SEOs can overlook during manual audits, ensuring every page has the schema type that best represents its primary purpose.

🛠️

Why Builders Need This

Your AI coding assistant doesn't just return data — it returns advice. When building a new site, the suggested schema gives you the correct type to implement from the start. No more shipping FAQPage when you should have shipped Service.

💰

No Extra Cost

The advisory engine is included in the standard $0.01 extraction fee for all MCP requests. There's no upsell. When your AI calls extract_schema, it automatically gets the advisory analysis whenever it could add value.

Setup in 60 Seconds

1. Generate an API key from the developer dashboard.
2. Top up your balance via Stripe ($10, $25, or $50).
3. Add the JSON block to your IDE's MCP config.
4. Ask your AI: "Use JSON Recon to extract the schema from example.com and write it into my layout."

Works with Claude Desktop, Cursor, Antigravity, Windsurf, and any IDE or platform that supports the Model Context Protocol.

Generate API Key →
claude_desktop_config.json
{
  "mcpServers": {
    "jsonrecon": {
      "command": "npx",
      "args": ["-y", "@jsonrecon/mcp-server"],
      "env": {
        "JSONRECON_API_KEY": "jr_liv_xxxxx"
      }
    }
  }
}
200+
Schema.org Entity Types
$0.01
Starting Per Extraction
Multi-Layer
Adaptive Extraction
~2s
Avg. Response Time
Auto
Refund on WAF Block