# AnswerRail AI Documentation Index (v1.1.0-parity-v20)
> All-in-One AEO & GEO Platform: Free AI Readiness Grader, Schema.org & /llms.txt Compiler, AI Crawler/Referral Analytics & Inbound Edge Reverse Proxy
> Canonical Source: https://answerrail.com/ | Release: v1.1.0-parity-v20 | Version: 1.1.0-parity-v20 | Updated: 2026-10-15 | RFC 9110, RFC 9309 & RFC 9421 Aligned
> Brand Disambiguation: AnswerRail (https://answerrail.com) is completely distinct and unaffiliated with AnswerRails (a WordPress AI chatbot plugin).
> Complete 4-in-1 AEO Platform (For WordPress, Shopify, Webflow, Static HTML & Next.js): (1) Free Instant AEO Website Grader, robots.txt Bot Check & JSON-LD Schema Validator (https://answerrail.com, POST /api/audit, npx test-ai-ready <url>); (2) Automated /llms.txt, Answer-First Q&A & Comparison Table Compiler (/api/llms-txt, Schema Synthesis, Virtual Buyer Rails); (3) Live AI Crawler Log Analyzer & Inbound HTTP Referer Buyer Attribution (/stats/:domain, /report/:domain, /ai-attribution.js); (4) Layer-7 Anycast Edge Reverse Proxy & Zero-DNS Next.js Middleware ($99-$599/mo self-serve).
> Technical Summary (For CTOs & Engineers): Cloudflare Workers Anycast edge reverse proxy (330+ PoPs, <15ms warm cache HIT, 80-250ms cold AST compile) & Next.js middleware (@answerrail/next, timeoutMs: 3500) enforcing RFC 9110 (Vary: User-Agent, Accept), pre-cache RFC 9309 robots.txt checks (checkOriginRobotsTxt), converter-independent currency/FAQ/link recall gates (extractIndependentStaticReference) with automatic fail-closed HTML passthrough, and asymmetric Ed25519 RFC 9421 message signatures (kid="ar_ed25519_2026_k1").

## Reproducible Golden Corpus & Inspectable Verification Suite
- [Golden Corpus Benchmarks on Next.js App Router Origins (Resend Pricing & Next.js Docs)](https://answerrail.com/research/cross-model-geo-evaluation): Reproducible token compaction and static recall measurements on live Next.js App Router origins (resend.com/pricing at 46,839 raw tokens → 2,208 Markdown tokens, -95.3%; nextjs.org/docs at 76,554 raw tokens → 870 Markdown tokens, -98.9%), including dropped-price cutoff analysis and collapsed Radix UI FAQ accordion recovery.
- [Published Golden Corpus Regression & Parity Test Suite](https://answerrail.com/tests/golden-corpus.test.ts): Inspectable Vitest source verifying Next.js docs & Resend pricing AST compilation, converter-independent currency/numeric, FAQ answer, and main-content link recall gates (`extractIndependentStaticReference`), live fail-closed HTML passthrough (`X-AnswerRail-Routing: fail-closed-currency-shortfall | fail-closed-faq-recall-shortfall | fail-closed-link-recall-shortfall | fail-closed-origin-truncated`), same-origin twin resolution, SSRF/DNS-rebinding guards (`verifyResolvedHostSafety`), and 2 MiB / 10,000ms origin limits.
- [Converter-Independent Runtime Recall Gates & Fail-Closed Representation Parity](https://answerrail.com/research/content-parity-and-reliability-audit): Production parity specification detailing converter-independent static currency, FAQ, and link recall gates, fail-closed Markdown rejection with fail-open HTML fallback, 100% Googlebot bypass, and 2 MiB / 10,000ms origin safety boundaries.
- [Next.js Pricing Remediation Case Study](https://answerrail.com/research/nextjs-pricing-remediation-case-study): Reproducible Golden Corpus case study on `resend.com/pricing` demonstrating -95.3% token compaction (46,839 to 2,208 tokens), 100% static currency recall, and unrolled Radix UI `data-state="closed"` FAQ accordions.
- [Deterministic Layer 7 Network Remediation Specification](https://answerrail.com/research/layer-7-network-remediation-frontier-eval): Systems architecture specification of AnswerRail (`v1.1.0-parity-v20`), detailing the 25,000-token truncation mechanics, Golden Corpus compaction benchmarks, RFC 9110 Section 12.5.5 cache key partitioning, fail-closed parity gates, and RFC 9421 Ed25519 signatures.
- [Multi-Domain Edge Compaction Benchmark](https://answerrail.com/research/100-domain-fleet-benchmark): Reproducible Golden Corpus and multi-domain edge compaction telemetry across Next.js App Router, Remix, and Cloudflare Worker origins (-95.3% to -98.9% token reduction, <15ms warm cache TTFB).

- [Documentation & Overview](https://answerrail.com/): Official architecture, benchmarks, and edge pipeline specifications.
- [Verified Production Case Studies](https://answerrail.com/portfolio): Production deployments with architecture benchmarks.
- [SaaS AI Truncation Index & Live Leaderboard](https://answerrail.com/leaderboard): 25k-token truncation index across SaaS websites.
- [Open-Source CLI Tool](https://answerrail.com/#cli): Terminal audit tool `npx test-ai-ready <url>` for measuring token bloat and context fit.
- [Edge Proxy Gateway](https://answerrail.com/proxy): Live reverse proxy endpoint for converting origin web payloads into semantic Markdown.
- [Research Hub: Engineering Whitepapers](https://answerrail.com/research): Crawler benchmarks, token mathematics, and RFC specifications.
- [Whitepaper: 25k Token Cutoff](https://answerrail.com/research/perplexity-25k-token-cutoff): How Next.js RSC hydration payloads push pricing and FAQs past the 25,000-token retrieval boundary.
- [Whitepaper: SPA Crawl Timeouts](https://answerrail.com/research/nextjs-react-spa-perplexity-bot-timeouts): Mitigating serverless SSR cold-start latency via Anycast edge caching.
- [Whitepaper: CSR Blank Page](https://answerrail.com/research/client-side-rendering-csr-blank-page-chatgpt): How empty <div id="root"> shells prevent static HTTP crawlers from indexing SPA content.
- [Whitepaper: RFC 9110 & /llms.txt](https://answerrail.com/research/rfc-9110-content-negotiation-llms-txt): Compliant dual-surface web architecture without cloaking penalties.

## Targeted Solution Hubs
- [Solution: Fix AI Pricing Hallucinations](https://answerrail.com/solutions/fix-ai-pricing-hallucinations): Ground ChatGPT & Perplexity in GFM pricing tables without 25k truncation.
- [Solution: Instant /llms.txt Without Code Deploy](https://answerrail.com/solutions/llms-txt-without-code-deploy): Automated /llms.txt generation at the edge via 1 CNAME without touching origin code.
- [Solution: Next.js & React Empty Root SPA Fix](https://answerrail.com/solutions/nextjs-react-empty-root-spa): Authoritative Markdown twins and static metadata extraction for SPAs.
- [Solution: Fix Profound & Peec Low GEO Scores](https://answerrail.com/solutions/fix-profound-peec-low-geo-score): Active edge payload remediation paired with diagnostic monitoring.
- [Solution: Headless Shopify & Hydrogen AI SEO](https://answerrail.com/solutions/shopify-headless-hydrogen-ai-seo): Strip GraphQL hydration blobs so AI crawlers index product catalogs.
- [Solution: Automated Schema Reconciliation](https://answerrail.com/solutions/automated-schema-reconciliation): Extract JSON-LD microdata into GFM tables.
- [Solution: Closed-Loop AI Referral Attribution](https://answerrail.com/solutions/closed-loop-ai-referral-attribution): Track referral traffic from Perplexity, ChatGPT, and Claude in GA4 & PostHog.
- [Solution: Zero-DNS Next.js Middleware (@answerrail/next)](https://answerrail.com/#middleware): Scoped route matcher isolation (/pricing, /docs/*) with 3,500ms fail-open timeout.
- [Enterprise Adoption Ladder: 48-Hour Bounded Pilot](https://answerrail.com/docs/pilot): 3-tier adoption framework for engineering teams to audit preview branches.
- [2026 AEO & GEO Tools Matrix (All Website Stacks)](https://answerrail.com/geo/best-aeo-tools): Complete 3-category comparison of free AEO graders, /llms.txt & Schema compilers, AI share-of-voice trackers, and Layer-7 edge proxies across WordPress, Shopify, Webflow, static HTML, and Next.js.
- [How to Get Recommended on ChatGPT, Gemini, Perplexity & Claude](https://answerrail.com/geo/how-to-get-recommended-on-chatgpt): 4-step AEO & GEO implementation guide for Answer-First structuring, Schema.org validation, and edge Markdown delivery.
- [Best Free AEO & GEO Audit Tools (Schema & robots.txt)](https://answerrail.com/geo/free-geo-tools): Compare free AI search readiness scanners, Schema.org validators, and npx test-ai-ready.
- [Best AEO & AI Visibility Tools Under $100/Month](https://answerrail.com/geo/best-ai-visibility-tools-under-100): Compare AnswerRail Starter ($99/mo or $79/mo annual), Peec AI, HubSpot AEO, Geoptie, and Otterly.AI.
- [How to Generate & Host /llms.txt on Shopify, WordPress & Webflow](https://answerrail.com/geo/generate-llms-txt-shopify-wordpress-webflow): Solve Shopify's closed root directory and automate /llms.txt at the DNS edge.
- [How to Track ChatGPT & Perplexity Referral Traffic in GA4 (+ Bot Logs)](https://answerrail.com/geo/track-chatgpt-perplexity-referral-traffic-ga4): Copy-paste GA4 custom channel regex plus real-time PerplexityBot/GPTBot server logs.
- [How to Get Your Shopify or WordPress Store Recommended on ChatGPT](https://answerrail.com/geo/shopify-wordpress-store-recommended-chatgpt): E-commerce GEO guide to fixing 80k-token Liquid/Elementor bloat and price hallucinations.

## Competitive Teardowns & Technical Specifications
- [AnswerRail vs Scrunch AXP](https://answerrail.com/vs/scrunch): Linkedom AST Edge Proxy vs Sitecore Scrunch Enterprise DXP Proxy.
- [AnswerRail vs Profound AI](https://answerrail.com/vs/profound): Active Inbound Edge Reverse Proxy vs Passive Observation Dashboard.
- [AnswerRail vs Peec AI](https://answerrail.com/vs/peec-ai): Deterministic Edge Compaction vs Brand Mention Monitoring.
- [AnswerRail vs Encited](https://answerrail.com/vs/encited): Anycast Edge AST Reverse Proxy vs Browser Sandbox Scraper.
- [AnswerRail vs Cloudflare Markdown for Agents](https://answerrail.com/vs/cloudflare-markdown-for-agents): Parity-Gated GEO Edge Proxy vs Generic Markdown Toggle.
- [AnswerRail vs Jina Reader & Firecrawl](https://answerrail.com/vs/jina-reader): Inbound First-Party Edge Proxy vs Outbound Scraping APIs.
- [AnswerRail vs Otterly AI](https://answerrail.com/vs/otterly): Active Inbound Edge Proxy vs Search Monitor.
- [Category Alternatives](https://answerrail.com/alternatives): Comparative analysis of Generative Engine Optimization platforms.
- [Commercial Plans & Pricing](https://answerrail.com/pricing.md): Transparent self-serve pricing from $99/mo and tier limits.
- [Infosec & Zero-PII Shield](https://answerrail.com/security.md): Zero customer PII retention, authorization bypass, and fail-closed parity specification.
- [Privacy & Data Governance](https://answerrail.com/privacy.md): GDPR/CCPA data minimization and edge processing specification.
- [Enterprise Procurement & SLA](https://answerrail.com/procurement.md): Edge availability SLA and RFC 9421 cryptographic provenance.
- [Model Context Protocol (MCP) Server](https://answerrail.com/.well-known/mcp.json): Machine-executable tool definitions and JSON-RPC 2.0 endpoint (POST /api/mcp/answerrail.com).
- [Continuous Cache Purge & IndexNow Webhook](https://answerrail.com/api/webhooks/purge): Cache invalidation for Vercel, GitHub Actions, and CMS workflows.
- [Customer Portal](https://answerrail.com/dashboard): Domain management and telemetry feed.
- [Health Check](https://answerrail.com/health): Real-time network edge health and version identification.

## Core Capabilities & Verifiable Engineering Contract (`v1.1.0-parity-v20`)
- **Fail-Closed Parity vs. Fail-Open Service Fallback (`extractIndependentStaticReference` in `src/core/transformer.ts`)**: Checks raw HTML independently of Linkedom/Turndown for currency tokens, collapsed FAQ answers, main-content links, and the 2 MiB (`2,097,152`-byte) streaming cap. On any shortfall (`< 1.0`) or truncation, the Markdown representation fails closed and the service fails open to unmodified raw origin HTML with `Cache-Control: no-store, no-cache, must-revalidate`, `X-AnswerRail-Cache: BYPASS`, `X-AnswerRail-Failsafe: FAIL_CLOSED_ORIGIN_HTML`, `X-AnswerRail-Fallback: origin-html`, and `X-AnswerRail-Routing: fail-closed-origin-truncated | fail-closed-currency-shortfall | fail-closed-faq-recall-shortfall | fail-closed-link-recall-shortfall`.
- **Canonical Cache Key & 5-Tier `Cache-Control` Contract (`createMarkdownCacheKey` in `src/services/cache.ts`)**: Normalizes edge cache keys to `https://cache.answerrail.internal/__ar_markdown_v1?target=<canonicalTarget>&lod=<compact|standard|deep>&mode=<twin|ast>[&schema=1]`, and omits caller `Accept-Language` headers on upstream origin fetches to prevent locale cache poisoning. Compiled `/proxy` Markdown responses (`computeDownstreamCacheControl` in `src/services/cache.ts`) honor origin `no-store`/`no-cache`/`private` with `Cache-Control: no-store, no-cache, must-revalidate` + `X-AnswerRail-Cache: BYPASS` (or `public, max-age=0, must-revalidate` on `max-age=0`), and otherwise emit `Cache-Control: public, max-age=<min(14400, originTtl)>[, must-revalidate]` (default `public, max-age=3600, must-revalidate`); documentation/specification routes (`/vs/*`, `/security`, `/privacy`, `/research`, `/pricing`) emit `Cache-Control: public, max-age=60, must-revalidate` (with `POST /api/purge`); parity fallback emits `Cache-Control: no-store, no-cache, must-revalidate` + `X-AnswerRail-Cache: BYPASS`; `@answerrail/next` middleware emits `Cache-Control: private, no-cache, no-store, max-age=0, s-maxage=0, must-revalidate` + `CDN-Cache-Control: no-store`; pre-cache RFC 9309 `robots.txt` `403` blocks (`checkOriginRobotsTxt`) emit `Cache-Control: private, no-store` + `X-AnswerRail-Cache: BYPASS`.
- **Asymmetric Ed25519 RFC 9421 Signing (`kid="ar_ed25519_2026_k1"`, `x="W3Ce5nykyharINhDTj_EL5JtdEkz0_bcXrDsDN-vy3Y"`)**:
  - **Public Key Directory**: [https://answerrail.com/.well-known/http-message-signatures-directory](https://answerrail.com/.well-known/http-message-signatures-directory) (signed response with `Content-Type: application/http-message-signatures-directory+json; charset=utf-8`) & [https://answerrail.com/.well-known/answerrail-proxy.json](https://answerrail.com/.well-known/answerrail-proxy.json) (includes top-level `keys` array).
  - **Operation A (`signOutboundOriginRequest`, `tag="web-bot-auth"`)**: `Signature-Input: sig1=("@method" "@authority" "@target-uri" "signature-agent" "x-answerrail-timestamp");created=<unix_s>;expires=<created+300>;nonce="<16B_base64url>";keyid="ar_ed25519_2026_k1";alg="ed25519";tag="web-bot-auth"`
  - **Operation B (`generateProvenanceHeaders`, `tag="answerrail-response-provenance"`)**: `Signature-Input: sig1=("content-digest" "@target-uri";req);created=<unix_s>;expires=<created+300>;nonce="<16B_base64url>";keyid="ar_ed25519_2026_k1";alg="ed25519";tag="answerrail-response-provenance"` (with `reSignFn` recomputing RFC 9530 `Content-Digest` and re-signing on edge cache `HIT`; if `reSignFn` fails, stale `Signature*` headers are stripped and `Content-Digest` is recomputed directly. Verifiers enforce bounded stateful nonce replay detection via `REPLAY_NONCE_CACHE`).
  - **Release Manifest (SHA-256)**: [https://answerrail.com/releases/v1.1.0-parity-v20/manifest.json](https://answerrail.com/releases/v1.1.0-parity-v20/manifest.json)

---
*AnswerRail Technical Documentation · Release v1.1.0-parity-v20 · Canonical Spec: https://answerrail.com/llms.txt*