seo: emit FAQPage JSON-LD on /ai/, /platform/, /pricing/; repoint /mcp/ links to /ai/ (#18)

GSC follow-ups from this week's coverage review:

1) Search Appearance was empty because no page emits rich-result-eligible
   structured data, even though /ai/, /platform/, and /pricing/ all have
   FAQ sections. Layout.astro now accepts an optional `structuredData`
   prop (object or array). The three FAQ pages declare their Q/A array
   in frontmatter and pass a FAQPage block. Default SoftwareApplication
   is now only emitted on the homepage instead of every non-article
   page, which was the source of the duplicate-entity warnings.

2) Two blog posts still linked to the retired /mcp/ path, which
   meta-refresh-redirects to /ai/. Repointed directly to /ai/ so the
   PageRank flow lands on the canonical target without the redirect
   hop:
   - src/content/blog/what-is-an-mcp-gateway.md:138
   - src/content/blog/why-agents-break-where-developers-cope.md:130

Layout type prop is still respected for blog posts (BlogPosting). Built
locally; FAQPage / Question / Answer triples emit on the three target
pages, SoftwareApplication only at /, BlogPosting only on articles.

Author: ndreno

src/content/blog/what-is-an-mcp-gateway.md

@@ -135,4 +135,4 @@ The MCP gateway category is in the same place the API gateway category was in ar
 
 If you are on a platform team and agents are not yet in your threat model, they are about to be. If you are on an AI product team and your agents are still calling internal APIs with hand-rolled tool servers, the shadow stack is already forming. The good news is the fix is cheap compared to the sprawl it prevents.
 
-At Barbacane we are building one gateway that spans all three categories: API gateway fundamentals, outbound AI gateway for your LLM calls (via the `ai-proxy` dispatcher), and MCP gateway for agents calling your APIs. Open source, Rust-native, spec-first, composed from the same dispatcher and middleware primitives so the three layers share plumbing rather than compete with it. If that sounds like what you need, [the /mcp page](/mcp/) is the five-minute version of the MCP layer specifically. Otherwise, the ideas above hold regardless of which tool you pick. The categories matter more than the product.
+At Barbacane we are building one gateway that spans all three categories: API gateway fundamentals, outbound AI gateway for your LLM calls (via the `ai-proxy` dispatcher), and MCP gateway for agents calling your APIs. Open source, Rust-native, spec-first, composed from the same dispatcher and middleware primitives so the three layers share plumbing rather than compete with it. If that sounds like what you need, [the /ai page](/ai/) is the five-minute version of the MCP layer specifically. Otherwise, the ideas above hold regardless of which tool you pick. The categories matter more than the product.

src/content/blog/why-agents-break-where-developers-cope.md

@@ -127,4 +127,4 @@ At Barbacane we build that platform on the assumption that this is where things
 
 ---
 
-*Barbacane is open source (AGPLv3) and available at [github.com/barbacane-dev/barbacane](https://github.com/barbacane-dev/barbacane). If MCP and agent-readiness are on your roadmap, [the /mcp page](/mcp/) is the short version of how we approach it.*
+*Barbacane is open source (AGPLv3) and available at [github.com/barbacane-dev/barbacane](https://github.com/barbacane-dev/barbacane). If MCP and agent-readiness are on your roadmap, [the /ai page](/ai/) is the short version of how we approach it.*

src/layouts/Layout.astro

@@ -11,6 +11,7 @@ interface Props {
     tags: string[];
   };
   ogImage?: string;
+  structuredData?: Record<string, unknown> | Record<string, unknown>[];
 }
 
 const {
@@ -19,13 +20,14 @@ const {
   type = 'website',
   article,
   ogImage: customOgImage,
+  structuredData: customStructuredData,
 } = Astro.props;
 
 const siteUrl = 'https://barbacane.dev';
 const canonicalUrl = new URL(Astro.url.pathname, siteUrl).href;
 const ogImage = customOgImage || `${siteUrl}/og-image.png`;
 
-const structuredData = type === 'article' && article
+const defaultStructuredData = type === 'article' && article
   ? {
       "@context": "https://schema.org",
       "@type": "BlogPosting",
@@ -46,7 +48,8 @@ const structuredData = type === 'article' && article
       "mainEntityOfPage": { "@type": "WebPage", "@id": canonicalUrl },
       "keywords": article.tags.join(", "),
     }
-  : {
+  : Astro.url.pathname === '/'
+  ? {
       "@context": "https://schema.org",
       "@type": "SoftwareApplication",
       "name": "Barbacane",
@@ -56,7 +59,14 @@ const structuredData = type === 'article' && article
       "url": siteUrl,
       "offers": { "@type": "Offer", "price": "0", "priceCurrency": "USD" },
       "author": { "@type": "Organization", "name": "Barbacane", "url": siteUrl },
-    };
+    }
+  : null;
+
+const structuredDataBlocks = customStructuredData
+  ? (Array.isArray(customStructuredData) ? customStructuredData : [customStructuredData])
+  : defaultStructuredData
+  ? [defaultStructuredData]
+  : [];
 ---
 
 <!doctype html>
@@ -121,7 +131,9 @@ const structuredData = type === 'article' && article
     <meta name="twitter:image" content={ogImage} />
 
     <!-- JSON-LD Structured Data -->
-    <script type="application/ld+json" set:html={JSON.stringify(structuredData)} />
+    {structuredDataBlocks.map((block) => (
+      <script type="application/ld+json" set:html={JSON.stringify(block)} />
+    ))}
   </head>
   <body>
     <nav class="fixed top-0 left-0 right-0 z-50 nav-bg backdrop-blur-lg border-b border-themed">

src/pages/ai.astro

@@ -73,11 +73,53 @@ paths:
 `;
 
 const codeThemes = { light: 'github-light', dark: 'github-dark' };
+
+const faq = [
+  {
+    question: "How is this different from Portkey or LiteLLM?",
+    answer: "They are outbound-only AI gateways. Barbacane handles both directions plus API gateway fundamentals, from one spec, with one middleware chain. A full comparison is published at /blog/barbacane-vs-portkey-litellm/.",
+  },
+  {
+    question: "Which LLM providers are supported on the outbound side?",
+    answer: "OpenAI, Anthropic, and Ollama out of the box, with pinned upstream API versions. Any OpenAI-compatible server (vLLM, TGI, LocalAI, Azure OpenAI, self-hosted inference) works via provider: openai plus a custom base_url.",
+  },
+  {
+    question: "Which MCP transport is supported on the inbound side?",
+    answer: "JSON-RPC 2.0 over HTTP POST on /__barbacane/mcp. Session termination via DELETE on the same path.",
+  },
+  {
+    question: "Can I adopt one direction without the other?",
+    answer: "Yes. The two directions are independent. You can run Barbacane as an outbound-only AI proxy, an inbound-only MCP gateway, or both, depending on which x-barbacane-* extensions your spec uses.",
+  },
+  {
+    question: "Do the AI governance middlewares apply to both directions?",
+    answer: "Yes. ai-prompt-guard, ai-token-limit, ai-cost-tracker, and ai-response-guard compose around any dispatcher. The same four middlewares govern outbound LLM calls and inbound MCP tool calls, so your cost attribution, token caps, and scrubbing policies live in one place.",
+  },
+  {
+    question: "Do I have to rewrite my existing APIs?",
+    answer: "No. Add x-barbacane-* extensions to your existing OpenAPI spec. Barbacane compiles it into an artifact that speaks both regular HTTP and MCP to the same upstream services.",
+  },
+  {
+    question: "Does authentication pass through?",
+    answer: "On the inbound side, the Authorization header from the MCP request is forwarded into the internal dispatch; your jwt-auth, apikey-auth, or OAuth2 middleware validates it like any other request. On the outbound side, provider credentials stay in dispatcher config and never leave it.",
+  },
+];
+
+const faqStructuredData = {
+  "@context": "https://schema.org",
+  "@type": "FAQPage",
+  "mainEntity": faq.map((item) => ({
+    "@type": "Question",
+    "name": item.question,
+    "acceptedAnswer": { "@type": "Answer", "text": item.answer },
+  })),
+};
 ---
 
 <Layout
   title="AI gateway for MCP and LLMs - Barbacane"
   description="Barbacane is the open-source AI gateway that handles both directions. Outbound: route LLM calls through OpenAI, Anthropic, or Ollama with provider fallback and policy routing. Inbound: expose your APIs to AI agents as typed MCP tools. Rust-native, spec-first, composed from dispatchers and middlewares."
+  structuredData={faqStructuredData}
 >
   <!-- Hero -->
   <section class="section relative overflow-hidden">

src/pages/platform.astro

@@ -1,10 +1,52 @@
 ---
 import Layout from '../layouts/Layout.astro';
+
+const faq = [
+  {
+    question: "How is this different from Kong, Apigee, or Tyk?",
+    answer: "Those gateways ship with their own config DSL. Your OpenAPI spec is a parallel document you have to keep in sync with the gateway. Barbacane inverts that: the spec is the config. There is no second source of truth to drift from.",
+  },
+  {
+    question: "Does it work with my existing OpenAPI specs?",
+    answer: "Yes. Add x-barbacane-* extensions where you need gateway behaviour. Existing fields (security schemes, parameters, responses) are already interpreted. Specs stay portable to other tools.",
+  },
+  {
+    question: "Can I add custom logic?",
+    answer: "Yes, via WebAssembly plugins. Barbacane ships a full suite of official plugins and lets you write your own in any WASM-compiling language. Plugins are sandboxed, so a faulty plugin cannot crash the gateway.",
+  },
+  {
+    question: "How does it deploy?",
+    answer: "A compiled artifact is a single binary. Run it anywhere: Kubernetes, bare metal, on-prem VMs, edge locations, air-gapped environments. The control plane and data planes are independent, so data planes keep serving even if the control plane is offline.",
+  },
+  {
+    question: "AGPLv3. Can I use it commercially?",
+    answer: "Yes. Small teams and non-profits get a free commercial license. Larger organisations can buy a paid commercial license. Either way, every feature is open source, no gated enterprise tier. See /pricing/ for details.",
+  },
+  {
+    question: "Does it support event-driven systems?",
+    answer: "Yes. Barbacane reads AsyncAPI specs and bridges HTTP to Kafka, NATS, or AWS Lambda. The same spec-first, compile-safe workflow applies.",
+  },
+  {
+    question: "Does this extend to AI traffic?",
+    answer: "Yes. The same spec-first model covers outbound LLM routing (via the ai-proxy dispatcher) and inbound MCP for AI agents. Platform teams often use Barbacane for all three on the same gateway. See /ai/ for that side of the story.",
+  },
+];
+
+const faqStructuredData = {
+  "@context": "https://schema.org",
+  "@type": "FAQPage",
+  "mainEntity": faq.map((item) => ({
+    "@type": "Question",
+    "name": item.question,
+    "acceptedAnswer": { "@type": "Answer", "text": item.answer },
+  })),
+};
 ---
 
 <Layout
   title="Spec-first API gateway for platform teams - Barbacane"
   description="Barbacane turns your OpenAPI and AsyncAPI specs into your API gateway. No config drift, no reconciliation work, no hidden policy gaps. Rust-native, extensible via sandboxed WebAssembly plugins, control and data plane split."
+  structuredData={faqStructuredData}
 >
   <!-- Hero -->
   <section class="section relative overflow-hidden">

src/pages/pricing.astro

@@ -1,8 +1,45 @@
 ---
 import Layout from '../layouts/Layout.astro';
+
+const faq = [
+  {
+    question: "Is Barbacane really free?",
+    answer: "Yes. All features are available to everyone with no gating. The source code is published under AGPLv3, which is free to use as long as you comply with its copyleft terms. If you cannot comply with AGPLv3 (e.g. you want to run it as a managed service without publishing modifications), a commercial license is free for small teams (up to €1M ARR, 10 employees) and non-profits.",
+  },
+  {
+    question: "What if I do not need paid services?",
+    answer: "That is perfectly fine. Use Barbacane freely, contribute to the project if you can, and reach out when you need help. Our documentation, GitHub issues, and community channels are available to everyone.",
+  },
+  {
+    question: "Can you help with compliance requirements?",
+    answer: "Yes. Our consulting packages include compliance reviews for SOC 2, GDPR, HIPAA, and other frameworks. We help you configure Barbacane's logging, security policies, AI governance middlewares, and audit trails to meet your requirements.",
+  },
+  {
+    question: "Does this cover AI and MCP workloads?",
+    answer: "Yes. Barbacane's AI gateway capabilities (ai-proxy dispatcher, AI governance middlewares) and MCP gateway capabilities are part of the same software. The same licensing and services apply, whether you are using Barbacane for outbound LLM routing, inbound agent tool exposure, or as a full API gateway.",
+  },
+  {
+    question: "Do you offer on-premise deployment?",
+    answer: "Barbacane is designed to run anywhere: on-prem, in your cloud, at the edge, or air-gapped. We can help you design and implement the deployment architecture that fits your infrastructure and security requirements.",
+  },
+];
+
+const faqStructuredData = {
+  "@context": "https://schema.org",
+  "@type": "FAQPage",
+  "mainEntity": faq.map((item) => ({
+    "@type": "Question",
+    "name": item.question,
+    "acceptedAnswer": { "@type": "Answer", "text": item.answer },
+  })),
+};
 ---
 
-<Layout title="Barbacane pricing: open-source AI and MCP gateway, commercial license available" description="Barbacane is open source under AGPLv3. A commercial license is free for small teams and non-profits, paid for larger orgs. Same software either way - spec-first AI, MCP, and API gateway.">
+<Layout
+  title="Barbacane pricing: open-source AI and MCP gateway, commercial license available"
+  description="Barbacane is open source under AGPLv3. A commercial license is free for small teams and non-profits, paid for larger orgs. Same software either way - spec-first AI, MCP, and API gateway."
+  structuredData={faqStructuredData}
+>
   <!-- Hero -->
   <section class="section">
     <div class="container-custom">