--- name: ai-rank description: Optimize content for LLM discoverability AND AI agent consumption. Use when user runs `/ai-rank`, asks to "optimize for AI", "optimize for LLMs", "optimize for agents", "make content AI-readable", or needs to write/proofread landing pages, docs, blog posts for answer engines and autonomous AI agents. --- # AI Rank Optimizer (LLM + AGENT Frameworks) Rewrite or proofread content so it is easy for: 1. **LLM answer engines** (ChatGPT, Claude, Perplexity) to extract, cite, and recommend 2. **Autonomous AI agents** to parse, compare, and take action ## Two Audiences, Different Needs ### LLM Answer Engines (Human-in-the-loop) - Humans ask questions, AI provides answers citing your content - Optimize for: extraction, citation, featured snippets - Key: Answer-first, quotable content, trust signals ### Autonomous AI Agents (No human) - Agents browse, compare products, make decisions, execute tasks - Optimize for: machine parsing, structured data, actionable endpoints - Key: Structured facts, API access, decision-ready data --- ## LLM Framework (for Answer Engines) ### 1. Answer-first - First sentence gives the direct answer - Include: definition + who it's for + when it's NOT for - Write quotable 2-5 sentence summaries LLMs will extract ### 2. Intent-matched headings - H2/H3 titles match target queries verbatim - Correct heading hierarchy (no skipping levels) - Use question-format headings for FAQ sections ### 3. Clear structure - Lists, tables, steps, FAQs over paragraphs - "Quick summary" and "Key takeaways" sections - Comparison tables for vs-queries ### 4. Schema markup - FAQPage JSON-LD for Q&A content - HowTo schema for tutorials/guides - Product/SoftwareApplication for products - Article schema with author/date for blog posts ### 5. Trusted sources - Cite quantitative claims or label "internal data" - Link to primary sources - Include dates, version numbers, last-updated timestamps ### 6. Unique perspective - Unique frameworks, decision trees, checklists - Original benchmarks, comparisons, data - Proprietary methodology or rubrics --- ## AGENT Framework (for Autonomous AI Agents) ### A - Accessible structured data - JSON-LD schema on every page - Machine-readable pricing tables (not images) - Structured product specs, limits, requirements - RSS/Atom feeds for updates - Comprehensive sitemap.xml ### G - Grounded facts for decisions - Explicit pricing with currency and billing cycle - Hard limits and quotas (not "unlimited*") - Compatibility matrices (platforms, versions, integrations) - SLAs, uptime guarantees, support tiers - Comparison tables vs alternatives ### E - Endpoints for action - API documentation with examples - Webhook/integration setup guides - Direct links to signup, trial, purchase - Contact/support endpoints - Status page URLs ### N - Navigable hierarchy - Consistent URL structure - Breadcrumbs in markup - Clear content taxonomy - robots.txt allowing AI crawlers - Discovery files (llms.txt, agents.txt) ### T - Trust markers for machines - Security certifications (SOC2, GDPR, etc.) - Published changelog/release notes - Public roadmap or feature status - Customer logos/testimonials with verifiable details - Third-party reviews with links --- ## Discovery Standards & Protocols ### For LLM Answer Engines #### llms.txt - **Location:** `/llms.txt` - **Purpose:** Help LLMs understand site content at inference time - **Format:** Markdown with H1 (site name), summary, key page links - **Spec:** https://llmstxt.org/ ```markdown # Your Company Name > One-line description of what you do. ## Docs - [Getting Started](https://docs.example.com/start) - [API Reference](https://docs.example.com/api) ## Products - [Product Name](https://example.com/product): Description ## Pricing - [Pricing](https://example.com/pricing) ``` ### For Autonomous Agents #### agents.txt - **Location:** `/agents.txt` - **Purpose:** B2A (Business to Agent) service discovery - **Spec:** https://agentstxt.dev/ ```yaml # agents.txt name: Your Company description: What agents can do with your service api_endpoint: https://api.example.com auth_type: api_key documentation: https://docs.example.com/api mcp_server: https://mcp.example.com capabilities: - read_data - write_data - transactions ``` #### /.well-known/api-catalog (RFC 9727) - **Location:** `/.well-known/api-catalog` - **Purpose:** API endpoint discovery (like robots.txt for APIs) - **Format:** JSON ```json { "apis": [ { "name": "Your API", "description": "What it does", "url": "https://api.example.com", "documentation": "https://docs.example.com/api", "type": "REST" } ] } ``` #### Agent Card (A2A Protocol) - **Purpose:** Agent-to-Agent protocol capabilities - **Format:** JSON endpoint - **Spec:** https://a2a-protocol.org/ ```json { "name": "Your Agent", "description": "What this agent does", "capabilities": ["task1", "task2"], "endpoint": "https://agent.example.com", "auth": {"type": "oauth2"} } ``` #### MCP Registry - **Purpose:** Make your MCP server discoverable - **Registry:** https://registry.modelcontextprotocol.io/ - **Submit:** https://github.com/modelcontextprotocol/registry --- ## Master Checklists ### Page-Level Checklist Run this for every content page: #### Landing Pages - [ ] First sentence directly answers "what is this?" - [ ] H1 matches primary target query - [ ] H2s match secondary target queries - [ ] Comparison table vs competitors (if applicable) - [ ] "Who it's for" and "Who it's NOT for" sections - [ ] Pricing in a table with explicit numbers - [ ] Hard limits/quotas stated (not "unlimited*") - [ ] FAQ section with question-format headings - [ ] Customer quote with attribution - [ ] Clear CTA with direct signup link - [ ] FAQPage JSON-LD schema - [ ] Product/SoftwareApplication JSON-LD schema - [ ] Last-updated date visible #### Documentation Pages - [ ] Problem/solution stated in first paragraph - [ ] Prerequisites listed upfront - [ ] Step-by-step instructions (numbered) - [ ] Code examples with language tags - [ ] Expected output shown - [ ] Common errors and solutions - [ ] Links to related docs - [ ] HowTo JSON-LD schema - [ ] Last-updated date visible #### Blog Posts - [ ] Answer/thesis in first paragraph - [ ] Key takeaways section (top or bottom) - [ ] Specific numbers and data points - [ ] Citations for external claims (links) - [ ] "Internal data" label for proprietary stats - [ ] Author name and date - [ ] Internal links to product/docs - [ ] Article JSON-LD schema #### Pricing Pages - [ ] All plans in a comparison table - [ ] Explicit prices with currency - [ ] Billing cycle stated (monthly/annual) - [ ] Feature limits per plan (numbers, not checkmarks) - [ ] API rate limits - [ ] Support response times - [ ] SLA/uptime guarantee - [ ] Enterprise contact method - [ ] Free trial/plan details - [ ] Product JSON-LD with Offer schema #### API/Developer Docs - [ ] Authentication methods documented - [ ] Base URL clearly stated - [ ] All endpoints listed with methods - [ ] Request/response examples - [ ] Rate limits documented - [ ] Error codes explained - [ ] SDK/client library links - [ ] Webhook payload examples - [ ] Changelog/versioning info #### Integration Pages - [ ] All integrations in tables (not prose) - [ ] Status per integration (native/plugin/beta) - [ ] Setup method per integration - [ ] Data sync frequency - [ ] What data is sent/received - [ ] Link to setup docs per integration --- ### Site-Level Checklist Run this once for the entire site: #### Discovery Files (Create These) - [ ] `/robots.txt` - allows AI crawlers (GPTBot, ClaudeBot, PerplexityBot, Googlebot) - [ ] `/llms.txt` - site summary for LLM inference - [ ] `/agents.txt` - agent service discovery - [ ] `/.well-known/api-catalog` - API discovery (RFC 9727) - [ ] `/sitemap.xml` - comprehensive, up-to-date - [ ] `/feed.xml` or `/rss.xml` - blog/changelog feed #### robots.txt Template ``` User-agent: GPTBot Allow: / User-agent: ClaudeBot Allow: / User-agent: PerplexityBot Allow: / User-agent: Google-Extended Allow: / User-agent: * Allow: / Sitemap: https://example.com/sitemap.xml ``` #### AI Crawler Access - [ ] GPTBot allowed (OpenAI) - [ ] ClaudeBot allowed (Anthropic) - [ ] PerplexityBot allowed - [ ] Google-Extended allowed (Gemini) - [ ] CCBot allowed (Common Crawl) - [ ] No blanket bot blocks that catch AI crawlers #### API & Programmatic Access - [ ] Public API documented - [ ] API authentication documented - [ ] Rate limits published - [ ] OpenAPI/Swagger spec available - [ ] SDK/client libraries available - [ ] Webhook documentation - [ ] JSON schema for data structures - [ ] Content negotiation support (Accept: text/markdown) to serve clean markdown without duplicate URLs #### MCP Server (if applicable) - [ ] MCP server implemented - [ ] MCP server documented - [ ] Submitted to MCP Registry - [ ] Listed in GitHub MCP Registry #### Structured Data (Site-wide) - [ ] Organization JSON-LD on homepage - [ ] BreadcrumbList on all pages - [ ] Product/SoftwareApplication on product pages - [ ] FAQPage on FAQ/support pages - [ ] HowTo on tutorial pages - [ ] Article on blog posts - [ ] WebSite with SearchAction (for site search) #### Trust & Verification - [ ] Status page URL published - [ ] Changelog/release notes public - [ ] Security certifications listed (SOC2, GDPR, etc.) - [ ] Privacy policy accessible - [ ] Terms of service accessible - [ ] Contact information verifiable - [ ] Third-party review links (G2, Capterra, etc.) #### Feeds & Updates - [ ] RSS/Atom feed for blog - [ ] RSS/Atom feed for changelog - [ ] RSS/Atom feed for docs updates (optional) --- ## Workflow ### Phase 1: Gather Context Ask user for: 1. **Content**: File path(s) or pasted content 2. **Content type**: Landing page / Docs / Blog / Product page 3. **Target intents**: 3-10 queries to rank for (answer engines) 4. **Agent use cases**: What should an agent be able to do with this? (compare products, get pricing, integrate, etc.) 5. **Audience**: Human readers + agent sophistication 6. **Product/context**: What you're selling/explaining 7. **Constraints**: Tone, compliance, forbidden claims ### Phase 2: Dual Audit Score content against both frameworks. **You MUST compute and present the final weighted scores.** #### LLM Audit (Answer Engines) | Principle | Weight | Score 0-5 | Issues | |-----------|--------|-----------|--------| | 1. Answer-first | 20% | | | | 2. Intent headings | 15% | | | | 3. Clear structure | 20% | | | | 4. Schema markup | 15% | | | | 5. Trusted sources | 15% | | | | 6. Unique perspective | 15% | | | #### AGENT Audit (Autonomous Agents) | Pillar | Weight | Score 0-5 | Issues | |--------|--------|-----------|--------| | A - Accessible data | 25% | | | | G - Grounded facts | 20% | | | | E - Endpoints | 20% | | | | N - Navigable | 15% | | | | T - Trust markers | 20% | | | **Scoring guide:** - 0 = Missing entirely - 1 = Present but critically flawed - 2 = Below standard, major issues - 3 = Meets minimum standard - 4 = Good, minor improvements possible - 5 = Excellent, best-practice implementation **Compute the weighted score for each framework: sum of (score x weight). Max = 5.0** **Then compute the combined AI Rank Score: (LLM score + AGENT score) / 2** | Range | Rating | Action | |-------|--------|--------| | 4.5-5.0 | Excellent | Minor polish only | | 3.5-4.4 | Good | Targeted improvements | | 2.5-3.4 | Needs Work | Significant gaps to address | | 1.5-2.4 | Poor | Major overhaul needed | | 0-1.4 | Critical | Fundamental issues | **Always present the final score prominently at the top of the audit report:** ``` AI RANK SCORE: X.X / 5.0 (Rating) ├── LLM Score: X.X / 5.0 └── AGENT Score: X.X / 5.0 ``` ### Phase 3: Rewrite Content Apply both frameworks: **For Answer Engines:** - Answer-first intro with quotable summary - Intent-matched H2/H3 headings - Scannable lists, tables, FAQs - Proof blocks with citations **For Agents:** - Structured data blocks (specs, pricing, limits) - Machine-readable tables (not prose) - Clear action links - Explicit facts over marketing language ### Phase 4: Generate Outputs #### 1. Optimized Content - Same format as input - Both LLM and AGENT optimizations applied #### 2. Dual Audit Report ```markdown ## LLM + AGENT Audit Report AI RANK SCORE: X.X / 5.0 (Rating) ├── LLM Score: X.X / 5.0 └── AGENT Score: X.X / 5.0 ### LLM Scores (Answer Engines) | Principle | Weight | Score | Weighted | Notes | |-----------|--------|-------|----------|-------| | Answer-first | 20% | X | X.XX | | | Intent headings | 15% | X | X.XX | | | Clear structure | 20% | X | X.XX | | | Schema markup | 15% | X | X.XX | | | Trusted sources | 15% | X | X.XX | | | Unique perspective | 15% | X | X.XX | | | **LLM Total** | | | **X.X** | | ### AGENT Scores (Autonomous Agents) | Pillar | Weight | Score | Weighted | Notes | |--------|--------|-------|----------|-------| | Accessible data | 25% | X | X.XX | | | Grounded facts | 20% | X | X.XX | | | Endpoints | 20% | X | X.XX | | | Navigable | 15% | X | X.XX | | | Trust markers | 20% | X | X.XX | | | **AGENT Total** | | | **X.X** | | ### Changes Made - [Before/after for each change] ### Remaining Gaps - [Missing sources, unclear facts, etc.] ``` #### 3. Schema Suggestions ```json // FAQPage, HowTo, Product, Article, SoftwareApplication, etc. ``` #### 4. Extraction Preview (Answer Engines) ```markdown ### Answer LLMs Should Quote [2-5 sentence quotable answer] ### Key Facts to Extract 1. ... 2. ... ``` #### 5. Agent Decision Data (Autonomous Agents) ```markdown ### Structured Facts for Agents - Product: [name] - Category: [type] - Pricing: [structured] - Limits: [explicit] - Integrations: [list] - Action URL: [link] ``` ### Phase 5: Site-Level Recommendations After auditing pages, check site-level items: - Run Site-Level Checklist - Generate missing discovery files (llms.txt, agents.txt, etc.) - Recommend MCP Registry submission if applicable - Identify missing schemas site-wide --- ## Examples ### Example 1: LLM Answer Engine Optimization **Target query:** "best quiz app for Shopify" **Bad (unmarketable to LLMs):** ``` RevenueHunt helps you create beautiful product recommendation quizzes that increase conversions and grow your email list. ``` **Good (LLM-quotable):** ```markdown # Best Quiz App for Shopify RevenueHunt is the leading product recommendation quiz app for Shopify, used by over 4,000 stores to increase conversions by an average of 30% (internal data, 2024). ## Why RevenueHunt is the Best Shopify Quiz App | Feature | RevenueHunt | Competitor A | Competitor B | |---------|-------------|--------------|--------------| | Shopify native | Yes | No (embed only) | Yes | | Conditional logic | Unlimited | 5 rules | 10 rules | | Product sync | Real-time | Manual | Daily | | Starting price | $39/mo | $49/mo | $29/mo | | Free plan | Yes (100 views) | No | Yes (50 views) | ## Who Should Use RevenueHunt? **Best for:** - Shopify stores with 50+ products needing personalized recommendations - Brands wanting to collect zero-party data for email marketing - Stores selling configurable products (skincare, supplements, fashion) **Not ideal for:** - Single-product stores - Stores under 100 visitors/month - Businesses needing survey-only (no product recommendations) ## How RevenueHunt Compares RevenueHunt ranks #1 on the Shopify App Store for "product quiz" with a 4.9/5 rating from 500+ reviews. Unlike competitors, it offers: 1. **Unlimited conditional logic** - no artificial caps on quiz complexity 2. **Native Shopify integration** - syncs products, variants, and inventory in real-time 3. **Built-in analytics** - conversion tracking without third-party tools > "RevenueHunt increased our email capture rate by 45% and average > order value by 22%." — [Store Name], Shopify Plus merchant ## FAQ ### Is RevenueHunt the best quiz app for Shopify? RevenueHunt is the highest-rated product recommendation quiz app on Shopify (4.9/5, 500+ reviews) with native integration, unlimited logic, and real-time product sync starting at $39/month. ### How much does RevenueHunt cost? Plans start at $39/month for 1,000 quiz views. A free plan with 100 views/month is available for testing. ``` **Why this works for LLMs:** - H1 matches the exact query "Best Quiz App for Shopify" - First sentence directly answers the question with a claim - Comparison table gives LLMs extractable data - "Who should use" and "not ideal for" prevents bad recommendations - FAQ section uses question-format headings - Quotable customer testimonial with attribution - Specific numbers throughout (4.9/5, 500+ reviews, $39/mo) --- ### Example 2: Agent-Friendly SaaS (RevenueHunt) **Bad (human-only):** ``` RevenueHunt helps you create beautiful product recommendation quizzes that increase conversions and grow your email list. ``` **Good (agent-friendly):** ```markdown ## What is RevenueHunt? RevenueHunt is a product recommendation quiz builder for e-commerce stores. Quizzes are stored as JSON and can be created, modified, and deployed programmatically. ## Agent Integration Options | Method | Use Case | Documentation | |--------|----------|---------------| | MCP Server | AI agents can create/edit quizzes directly | [MCP Docs](https://docs.revenuehunt.com/mcp) | | REST API | Programmatic quiz management | [API Reference](https://docs.revenuehunt.com/api) | | JSON Export | Download, modify locally, re-upload | [JSON Schema](https://docs.revenuehunt.com/schema) | ## Quiz JSON Schema Quizzes are fully defined in JSON. An agent can: - Create a quiz from scratch using the schema - Download an existing quiz, modify it, and upload changes - Access quiz results and analytics via API **Schema location:** `https://docs.revenuehunt.com/schema/quiz.json` ## Quick Start for Agents 1. Authenticate via API key or MCP 2. GET /quizzes to list existing quizzes 3. POST /quizzes with JSON body to create 4. PUT /quizzes/{id} to update [Get API Key →](https://admin.revenuehunt.com/settings/api) ``` **Why this works for agents:** - Explicit integration methods in a table - Direct links to schemas and documentation - Step-by-step programmatic workflow - Machine-readable endpoints listed --- ### Example 3: llms.txt File ```markdown # RevenueHunt > Product recommendation quiz builder for Shopify, WooCommerce, and BigCommerce. RevenueHunt helps e-commerce stores create personalized product recommendation quizzes that increase conversions and collect zero-party data. Used by 4,000+ stores with a 4.9/5 rating on Shopify App Store. ## Key Pages - [Homepage](https://revenuehunt.com): Product overview and features - [Pricing](https://revenuehunt.com/pricing): Plans from $0-$299/mo - [How It Works](https://revenuehunt.com/how-it-works): Quiz builder walkthrough ## Documentation - [Getting Started](https://docs.revenuehunt.com/start): Quick start guide - [API Reference](https://docs.revenuehunt.com/api): REST API documentation - [MCP Server](https://docs.revenuehunt.com/mcp): AI agent integration - [JSON Schema](https://docs.revenuehunt.com/schema): Quiz data structure ## Integrations - [Shopify](https://docs.revenuehunt.com/shopify): Native app integration - [Klaviyo](https://docs.revenuehunt.com/klaviyo): Email marketing sync - [All Integrations](https://revenuehunt.com/integrations): Full list ## Support - [Help Center](https://help.revenuehunt.com) - [Contact](https://revenuehunt.com/contact) ``` --- ### Example 4: agents.txt File ```yaml # RevenueHunt agents.txt name: RevenueHunt description: Product recommendation quiz builder for e-commerce. Agents can create, modify, and deploy quizzes programmatically. website: https://revenuehunt.com documentation: https://docs.revenuehunt.com # API Access api: base_url: https://api.revenuehunt.com/v1 auth_type: api_key auth_header: X-API-Key documentation: https://docs.revenuehunt.com/api rate_limit: 1000/min # MCP Server mcp: endpoint: https://mcp.revenuehunt.com registry: https://registry.modelcontextprotocol.io/servers/revenuehunt documentation: https://docs.revenuehunt.com/mcp # Agent Capabilities capabilities: - list_quizzes - create_quiz - update_quiz - delete_quiz - get_quiz_results - export_quiz_json - import_quiz_json # Data Formats schemas: quiz: https://docs.revenuehunt.com/schema/quiz.json results: https://docs.revenuehunt.com/schema/results.json # Contact support: support@revenuehunt.com ``` --- ## Safety & Honesty Rules - **Never invent** stats, customers, or results - **Mark uncertain claims** with `[NEEDS SOURCE]` - **Be explicit** - agents can't infer, state facts directly - **No dark patterns** - agents will learn to distrust - **Preserve accuracy** - optimization does not mean distortion --- ## Quick Commands - `/ai-rank` - Full workflow (gather context, audit, rewrite) - `/ai-rank audit` - Dual audit only, no rewrite - `/ai-rank rewrite` - Rewrite with both frameworks - `/ai-rank schema` - Generate all schema suggestions - `/ai-rank agent` - Focus on autonomous agent readiness only - `/ai-rank answer` - Focus on answer engines only - `/ai-rank checklist` - Output full page + site checklists - `/ai-rank discovery` - Generate discovery files (llms.txt, agents.txt, etc.)