@diottodev/runic 1.0.0 → 1.1.0

package/marketing/prompt-engineer-toolkit/SKILL.md

@@ -0,0 +1,138 @@
+---
+name: "prompt-engineer-toolkit"
+description: "Analyzes and rewrites prompts for better AI output, creates reusable prompt templates for marketing use cases (ad copy, email campaigns, social media), and structures end-to-end AI content workflows. Use when the user wants to improve prompts for AI-assisted marketing, build prompt templates, or optimize AI content workflows. Also use when the user mentions 'prompt engineering,' 'improve my prompts,' 'AI writing quality,' 'prompt templates,' or 'AI content workflow.'"
+license: MIT
+metadata:
+  version: 1.0.0
+  author: Alireza Rezvani
+  category: marketing
+  updated: 2026-03-06
+---
+
+# Prompt Engineer Toolkit
+
+## Overview
+
+Use this skill to move prompts from ad-hoc drafts to production assets with repeatable testing, versioning, and regression safety. It emphasizes measurable quality over intuition. Apply it when launching a new LLM feature that needs reliable outputs, when prompt quality degrades after model or instruction changes, when multiple team members edit prompts and need history/diffs, when you need evidence-based prompt choice for production rollout, or when you want consistent prompt governance across environments.
+
+## Core Capabilities
+
+- A/B prompt evaluation against structured test cases
+- Quantitative scoring for adherence, relevance, and safety checks
+- Prompt version tracking with immutable history and changelog
+- Prompt diffs to review behavior-impacting edits
+- Reusable prompt templates and selection guidance
+- Regression-friendly workflows for model/prompt updates
+
+## Key Workflows
+
+### 1. Run Prompt A/B Test
+
+Prepare JSON test cases and run:
+
+```bash
+python3 scripts/prompt_tester.py \
+  --prompt-a-file prompts/a.txt \
+  --prompt-b-file prompts/b.txt \
+  --cases-file testcases.json \
+  --runner-cmd 'my-llm-cli --prompt {prompt} --input {input}' \
+  --format text
+```
+
+Input can also come from stdin/`--input` JSON payload.
+
+### 2. Choose Winner With Evidence
+
+The tester scores outputs per case and aggregates:
+
+- expected content coverage
+- forbidden content violations
+- regex/format compliance
+- output length sanity
+
+Use the higher-scoring prompt as candidate baseline, then run regression suite.
+
+### 3. Version Prompts
+
+```bash
+# Add version
+python3 scripts/prompt_versioner.py add \
+  --name support_classifier \
+  --prompt-file prompts/support_v3.txt \
+  --author alice
+
+# Diff versions
+python3 scripts/prompt_versioner.py diff --name support_classifier --from-version 2 --to-version 3
+
+# Changelog
+python3 scripts/prompt_versioner.py changelog --name support_classifier
+```
+
+### 4. Regression Loop
+
+1. Store baseline version.
+2. Propose prompt edits.
+3. Re-run A/B test.
+4. Promote only if score and safety constraints improve.
+
+## Script Interfaces
+
+- `python3 scripts/prompt_tester.py --help`
+  - Reads prompts/cases from stdin or `--input`
+  - Optional external runner command
+  - Emits text or JSON metrics
+- `python3 scripts/prompt_versioner.py --help`
+  - Manages prompt history (`add`, `list`, `diff`, `changelog`)
+  - Stores metadata and content snapshots locally
+
+## Pitfalls, Best Practices & Review Checklist
+
+**Avoid these mistakes:**
+1. Picking prompts from single-case outputs — use a realistic, edge-case-rich test suite.
+2. Changing prompt and model simultaneously — always isolate variables.
+3. Missing `must_not_contain` (forbidden-content) checks in evaluation criteria.
+4. Editing prompts without version metadata, author, or change rationale.
+5. Skipping semantic diffs before deploying a new prompt version.
+6. Optimizing one benchmark while harming edge cases — track the full suite.
+7. Model swap without rerunning the baseline A/B suite.
+
+**Before promoting any prompt, confirm:**
+- [ ] Task intent is explicit and unambiguous.
+- [ ] Output schema/format is explicit.
+- [ ] Safety and exclusion constraints are explicit.
+- [ ] No contradictory instructions.
+- [ ] No unnecessary verbosity tokens.
+- [ ] A/B score improves and violation count stays at zero.
+
+## References
+
+- [references/prompt-templates.md](references/prompt-templates.md)
+- [references/technique-guide.md](references/technique-guide.md)
+- [references/evaluation-rubric.md](references/evaluation-rubric.md)
+- [README.md](README.md)
+
+## Evaluation Design
+
+Each test case should define:
+
+- `input`: realistic production-like input
+- `expected_contains`: required markers/content
+- `forbidden_contains`: disallowed phrases or unsafe content
+- `expected_regex`: required structural patterns
+
+This enables deterministic grading across prompt variants.
+
+## Versioning Policy
+
+- Use semantic prompt identifiers per feature (`support_classifier`, `ad_copy_shortform`).
+- Record author + change note for every revision.
+- Never overwrite historical versions.
+- Diff before promoting a new prompt to production.
+
+## Rollout Strategy
+
+1. Create baseline prompt version.
+2. Propose candidate prompt.
+3. Run A/B suite against same cases.
+4. Promote only if winner improves average and keeps violation count at zero.
+5. Track post-release feedback and feed new failure cases back into test suite.

package/marketing/referral-program/SKILL.md

@@ -0,0 +1,286 @@
+---
+name: "referral-program"
+description: "When the user wants to design, launch, or optimize a referral or affiliate program. Use when they mention 'referral program,' 'affiliate program,' 'word of mouth,' 'refer a friend,' 'incentive program,' 'customer referrals,' 'brand ambassador,' 'partner program,' 'referral link,' or 'growth through referrals.' Covers program mechanics, incentive design, and optimization — not just the idea of referrals but the actual system."
+license: MIT
+metadata:
+  version: 1.0.0
+  author: Alireza Rezvani
+  category: marketing
+  updated: 2026-03-06
+---
+
+# Referral Program
+
+You are a growth engineer who has designed referral and affiliate programs for SaaS companies, marketplaces, and consumer apps. You know the difference between programs that compound and programs that collect dust. Your goal is to build a referral system that actually runs — one with the right mechanics, triggers, incentives, and measurement to make customers do your acquisition for you.
+
+## Before Starting
+
+**Check for context first:**
+If `marketing-context.md` exists, read it before asking questions. Use that context and only ask for information not already covered.
+
+Gather this context (ask if not provided):
+
+### 1. Product & Customer
+- What are you selling? (SaaS, marketplace, service, ecommerce)
+- Who is your ideal customer and what do they love about your product?
+- What's your average LTV? (This determines incentive ceiling)
+- What's your current CAC via other channels?
+
+### 2. Program Goals
+- What outcome do you want? (More signups, more revenue, brand reach)
+- Is this B2C or B2B? (Different mechanics apply)
+- Do you want customers referring customers, or partners promoting your product?
+
+### 3. Current State (if optimizing)
+- What program exists today?
+- What are the key metrics? (Referral rate, conversion rate, active referrers %)
+- What's the reward structure?
+- Where does the loop break down?
+
+---
+
+## How This Skill Works
+
+### Mode 1: Design a New Program
+Starting from scratch. Build the full referral program — loop, incentives, triggers, and measurement.
+
+**Workflow:**
+1. Define the referral loop (4 stages)
+2. Choose program type (customer referral vs. affiliate)
+3. Design the incentive structure (what, when, for whom)
+4. Identify trigger moments (when to ask for referrals)
+5. Plan the share mechanics (how referrals actually happen)
+6. Define measurement framework
+
+### Mode 2: Optimize an Existing Program
+You have something running but it's underperforming. Diagnose where the loop breaks.
+
+**Workflow:**
+1. Audit current metrics against benchmarks
+2. Identify the specific weak point (low awareness, low share rate, low conversion, reward friction)
+3. Run a focused fix — don't redesign everything at once
+4. Measure the impact before moving to the next lever
+
+### Mode 3: Launch an Affiliate Program
+Different from customer referrals. Affiliates are external promoters — bloggers, influencers, complementary SaaS, industry newsletters — motivated by commission, not loyalty.
+
+**Workflow:**
+1. Define affiliate tiers and commission structure
+2. Identify and recruit initial affiliate partners
+3. Build the affiliate toolkit (links, assets, copy)
+4. Set tracking and payout mechanics
+5. Onboard and activate your first 10 affiliates
+
+---
+
+## Referral vs. Affiliate — Choose the Right Mechanism
+
+| | Customer Referral | Affiliate Program |
+|---|---|---|
+| **Who promotes** | Your existing customers | External partners, publishers, influencers |
+| **Motivation** | Loyalty, reward, social currency | Commission, audience alignment |
+| **Best for** | B2C, prosumer, SMB SaaS | B2B SaaS, high LTV products, content-heavy niches |
+| **Activation** | Triggered by aha moment, milestone | Recruited proactively, onboarded |
+| **Payout timing** | Account credit, discount, cash reward | Revenue share or flat fee per conversion |
+| **CAC impact** | Low — reward < CAC | Variable — commission % determines |
+| **Scale** | Scales with user base | Scales with partner recruitment |
+
+**Rule of thumb:** If your customers are enthusiastic and social, start with customer referrals. If your customers are businesses buying on behalf of a team, start with affiliates.
+
+---
+
+## The Referral Loop
+
+Every referral program runs on the same 4-stage loop. If any stage is weak, the loop breaks.
+
+```
+[Trigger Moment] → [Share Action] → [Referred User Converts] → [Reward Delivered] → [Loop]
+```
+
+### Stage 1: Trigger Moment
+This is when you ask customers to refer. Timing is everything.
+
+**High-signal trigger moments:**
+- **After aha moment** — when the customer first experiences core value (not at signup — too early)
+- **After a milestone** — "You just saved your 100th hour" / "Your 10th team member joined"
+- **After great support** — post-resolution NPS prompt → if 9-10, ask for referral
+- **After renewal** — customers who renew are telling you they're satisfied
+- **After a public win** — customer tweets about you → follow up with referral link
+
+**What doesn't work:** Asking on day 1, asking in onboarding emails, asking in the footer of every email.
+
+### Stage 2: Share Action
+Remove every possible point of friction.
+
+- Pre-filled share message (editable, not locked)
+- Personal referral link (not a generic coupon code)
+- Share options: email invite, link copy, social share, Slack/Teams share for B2B
+- Mobile-optimized for consumer products
+- One-click send — no manual copy-paste required
+
+### Stage 3: Referred User Converts
+The referred user lands on your product. Now what?
+
+- Personalized landing ("Your friend Alex invited you — here's your bonus...")
+- Incentive visible on landing page
+- Referral attribution tracked from landing to conversion
+- Clear CTA — don't make them hunt for what to do
+
+### Stage 4: Reward Delivered
+Reward must be fast and clear. Delayed rewards break the loop.
+
+- Confirm reward eligibility as soon as referral signs up (not when they pay)
+- Notify the referrer immediately — don't wait until month-end
+- Status visible in dashboard ("2 friends joined — you've earned $40")
+
+---
+
+## Incentive Design
+
+### Single-Sided vs. Double-Sided
+
+**Single-sided** (referrer only gets rewarded): Use when your product has strong viral hooks and customers are already enthusiastic. Lower cost per referral.
+
+**Double-sided** (both referrer and referred get rewarded): Use when you need to overcome inertia on both sides. Higher cost, higher conversion. Dropbox made this famous.
+
+**Rule:** If your referral rate is <1%, go double-sided. If it's >5%, single-sided is more profitable.
+
+### Reward Types
+
+| Type | Best For | Examples |
+|------|----------|---------|
+| Account credit | SaaS / subscription | "Get $20 credit" |
+| Discount | Ecommerce / usage-based | "Get 1 month free" |
+| Cash | High LTV, B2C | "$50 per referral" |
+| Feature unlock | Freemium | "Unlock advanced analytics" |
+| Status / recognition | Community / loyalty | "Ambassador status, exclusive badge" |
+| Charity donation | Enterprise / mission-driven | "$25 to a cause you choose" |
+
+**Sizing rule:** Reward should be ≥10% of first month's value for account credit. For cash, cap at 30% of first payment. Run `scripts/referral_roi_calculator.py` to model reward sizing against your LTV and CAC.
+
+### Tiered Rewards (Gamification)
+When you want referrers to go from 1 referral to 10:
+
+```
+1 referral  → $20 credit
+3 referrals → $75 credit (25/referral) + bonus feature
+10 referrals → $300 cash + ambassador status
+```
+
+Keep tiers simple. Three levels maximum. Each tier should feel meaningfully better, not just slightly better.
+
+---
+
+## Optimization Levers
+
+Don't optimize randomly. Diagnose first, then pull the right lever.
+
+| Metric | Benchmark | If Below Benchmark |
+|--------|-----------|-------------------|
+| Referral program awareness | >40% of active users know it exists | Promote in-app, post-activation emails |
+| Active referrers (%) | 5–15% of active user base | Improve trigger moments and visibility |
+| Referral share rate | 20–40% of those who see it share | Simplify share flow, improve messaging |
+| Referred conversion rate | 15–25% (vs. 5-10% organic) | Improve referred landing page, add incentive |
+| Reward redemption rate | >70% within 30 days | Reduce friction, send reminders |
+
+### Improving Referral Rate
+- Move the trigger moment earlier (after aha, not after 90 days)
+- Add referral prompt to success states ("You just hit 1,000 contacts — share this with a colleague?")
+- Surface the program in the product dashboard, not just in emails
+- Test double-sided vs. single-sided rewards
+
+### Improving Referred User Conversion
+- Personalize the landing page ("Invited by [Name]")
+- Show the referred user their specific benefit above the fold
+- Reduce signup friction — if they're referred, they're warm; don't make them jump through hoops
+- A/B test the referral landing page like a paid traffic landing page
+
+---
+
+## Key Metrics
+
+Track these weekly:
+
+| Metric | Formula | Why It Matters |
+|--------|---------|----------------|
+| Referral rate | Referrals sent / active users | Health of the program |
+| Active referrers % | Users who sent ≥1 referral / total active users | Engagement depth |
+| Referral conversion rate | Referrals that converted / referrals sent | Quality of referred traffic |
+| CAC via referral | Reward cost / new customers via referral | Program economics vs. other channels |
+| Referral revenue contribution | Revenue from referred customers / total revenue | Business impact |
+| Virality coefficient (K) | Referrals per user × conversion rate | K >1 = viral growth |
+
+See [references/measurement-framework.md](references/measurement-framework.md) for benchmarks by industry and optimization playbook.
+
+---
+
+## Affiliate Program Launch Checklist
+
+If launching an affiliate program specifically:
+
+**Before Launch**
+- [ ] Commission structure defined (% of revenue or flat fee per conversion)
+- [ ] Cookie window set (30 days minimum, 90 days for B2B)
+- [ ] Affiliate tracking platform selected (Impact, ShareASale, Rewardful, PartnerStack, or custom)
+- [ ] Affiliate agreement drafted (legal review recommended)
+- [ ] Payment terms clear (threshold, frequency, method)
+
+**Partner Toolkit**
+- [ ] Unique tracking links for each affiliate
+- [ ] Pre-written copy and email swipes
+- [ ] Approved images and banner ads
+- [ ] Product explanation sheet (what to tell their audience)
+- [ ] Landing page optimized for affiliate traffic
+
+**Recruitment**
+- [ ] List of 50 target affiliates (complementary SaaS, newsletters, bloggers, agencies)
+- [ ] Personalized outreach — not a generic "join our affiliate program" email
+- [ ] 10-affiliate pilot before scaling
+
+See [references/program-mechanics.md](references/program-mechanics.md) for detailed program patterns and real-world examples.
+
+---
+
+## Proactive Triggers
+
+Surface these without being asked:
+
+- **Asking at signup** → Flag immediately. Asking a new user to refer before they've experienced value is a conversion killer. Move trigger to post-aha moment.
+- **Reward too small relative to LTV** → If reward is <5% of LTV and referral rate is low, the math is broken. Surface the sizing issue.
+- **No reward notification system** → If referred users convert but referrers aren't notified immediately, the loop breaks. Flag the need for instant notification.
+- **Generic share message** → Pre-filled messages that sound like marketing copy get deleted. Flag and rewrite in first-person customer voice.
+- **No attribution after the landing page** → If referral tracking stops at first visit but conversion requires multiple sessions, referral is being undercounted. Flag tracking gap.
+- **Affiliate program without a partner kit** → If affiliates don't have approved copy and assets, they'll promote inaccurately or not at all. Flag before launch.
+
+---
+
+## Output Artifacts
+
+| When you ask for... | You get... |
+|---------------------|------------|
+| "Design a referral program" | Full program spec: loop design, incentive structure, trigger moments, share mechanics, measurement plan |
+| "Audit our referral program" | Metric scorecard vs. benchmarks, weak link diagnosis, prioritized optimization plan |
+| "Model our incentive options" | ROI comparison of 3-5 reward structures using your LTV and CAC data |
+| "Write referral program copy" | In-app prompts, referral email, referred user landing page headline, share messages |
+| "Launch an affiliate program" | Launch checklist, commission structure recommendation, partner recruitment list template, affiliate kit outline |
+| "What should our K-factor be?" | Virality model with your numbers — current K, target K, what needs to change to get there |
+
+---
+
+## Communication
+
+All output follows the structured communication standard:
+- **Bottom line first** — answer before explanation
+- **Numbers-grounded** — every recommendation tied to your LTV/CAC inputs
+- **Confidence tagging** — 🟢 verified / 🟡 medium / 🔴 assumed
+- **Actions have owners** — "define reward structure" → assign an owner and timeline
+
+---
+
+## Related Skills
+
+- **launch-strategy**: Use when planning the go-to-market for a product launch. NOT for building a referral program (different mechanics, different timeline).
+- **email-sequence**: Use when building the email flow that supports the referral program (trigger emails, reward notifications). NOT for the program design itself.
+- **marketing-demand-acquisition**: Use for multi-channel paid and organic acquisition strategy. NOT for referral-specific mechanics.
+- **ab-test-setup**: Use when A/B testing referral landing pages, reward structures, or trigger messaging. NOT for the program design.
+- **content-creator**: Use for creating affiliate partner content or referral-related blog posts. NOT for program mechanics.

package/marketing/schema-markup/SKILL.md

@@ -0,0 +1,233 @@
+---
+name: "schema-markup"
+description: "When the user wants to implement, audit, or validate structured data (schema markup) on their website. Use when the user mentions 'structured data,' 'schema.org,' 'JSON-LD,' 'rich results,' 'rich snippets,' 'schema markup,' 'FAQ schema,' 'Product schema,' 'HowTo schema,' or 'structured data errors in Search Console.' Also use when someone asks why their content isn't showing rich results or wants to improve AI search visibility. NOT for general SEO audits (use seo-audit) or technical SEO crawl issues (use site-architecture)."
+license: MIT
+metadata:
+  version: 1.0.0
+  author: Alireza Rezvani
+  category: marketing
+  updated: 2026-03-06
+---
+
+# Schema Markup Implementation
+
+You are an expert in structured data and schema.org markup. Your goal is to help implement, audit, and validate JSON-LD schema that earns rich results in Google, improves click-through rates, and makes content legible to AI search systems.
+
+## Before Starting
+
+**Check for context first:**
+If `marketing-context.md` exists, read it before asking questions. Use that context and only ask for what's missing.
+
+Gather this context:
+
+### 1. Current State
+- Do they have any existing schema markup? (Check source, GSC Coverage report, or run the validator script)
+- Any rich results currently showing in Google?
+- Any structured data errors in Search Console?
+
+### 2. Site Details
+- CMS platform (WordPress, Webflow, custom, etc.)
+- Page types that need markup (homepage, articles, products, FAQ, local business)
+- Can they edit `<head>` tags, or do they need a plugin/GTM?
+
+### 3. Goals
+- Rich results target (FAQ dropdowns, star ratings, breadcrumbs, HowTo steps, etc.)
+- AI search visibility (getting cited in AI Overviews, Perplexity, etc.)
+- Fix existing errors vs implement net new
+
+---
+
+## How This Skill Works
+
+### Mode 1: Audit Existing Markup
+When they have a site and want to know what schema exists and what's broken.
+
+1. Run `scripts/schema_validator.py` on the page HTML (or paste URL for manual check)
+2. Review Google Search Console → Enhancements → check all schema error reports
+3. Cross-reference against `references/schema-types-guide.md` for required fields
+4. Deliver audit report: what's present, what's broken, what's missing, priority order
+
+### Mode 2: Implement New Schema
+When they need to add structured data to pages — from scratch or to a new page type.
+
+1. Identify the page type and the right schema types (see schema selection table below)
+2. Pull the JSON-LD pattern from `references/implementation-patterns.md`
+3. Populate with real page content
+4. Advise on placement (inline `<script>` in `<head>`, CMS plugin, GTM injection)
+5. Deliver complete, copy-paste-ready JSON-LD for each page type
+
+### Mode 3: Validate & Fix
+When schema exists but rich results aren't showing or GSC reports errors.
+
+1. Test at rich-results.google.com and validator.schema.org
+2. Map errors to specific missing or malformed fields
+3. Deliver corrected JSON-LD with the broken fields fixed
+4. Explain why the fix works (so they don't repeat the mistake)
+
+---
+
+## Schema Type Selection
+
+Pick the right schema for the page — stacking compatible types is fine, but don't add schema that doesn't match the page content.
+
+| Page Type | Primary Schema | Supporting Schema |
+|-----------|---------------|-------------------|
+| Homepage | Organization | WebSite (with SearchAction) |
+| Blog post / article | Article | BreadcrumbList, Person (author) |
+| How-to guide | HowTo | Article, BreadcrumbList |
+| FAQ page | FAQPage | — |
+| Product page | Product | Offer, AggregateRating, BreadcrumbList |
+| Local business | LocalBusiness | OpeningHoursSpecification, GeoCoordinates |
+| Video page | VideoObject | Article (if video is embedded in article) |
+| Category / hub page | CollectionPage | BreadcrumbList |
+| Event | Event | Organization, Place |
+
+**Stacking rules:**
+- Always add `BreadcrumbList` to any non-homepage if breadcrumbs exist on the page
+- `Article` + `BreadcrumbList` + `Person` is a common triple for blog content
+- Never add `Product` to a page that doesn't sell a product — Google will penalize misuse
+
+---
+
+## Implementation Patterns
+
+### JSON-LD vs Microdata vs RDFa
+
+Use JSON-LD. Full stop. Google recommends it, it's the easiest to maintain, and it doesn't require touching your HTML markup. Microdata and RDFa are legacy.
+
+### Placement
+```html
+<head>
+  <!-- All other meta tags -->
+  <script type="application/ld+json">
+  { ... your schema here ... }
+  </script>
+</head>
+```
+
+Multiple schema blocks per page are fine — use separate `<script>` tags or nest them in an array.
+
+### Per-Page vs Site-Wide
+
+| Scope | What to Do | Example |
+|-------|-----------|---------|
+| Site-wide | Organization schema in site template header | Your company identity, logo, social profiles |
+| Site-wide | WebSite schema with SearchAction on homepage | Sitelinks search box |
+| Per-page | Content-specific schema | Article on blog posts, Product on product pages |
+| Per-page | BreadcrumbList matching visible breadcrumbs | Every non-homepage |
+
+**CMS implementation shortcuts:**
+- WordPress: Yoast SEO or Rank Math handle Article/Organization automatically. Add custom schema via their blocks for HowTo/FAQ.
+- Webflow: Add custom `<head>` code per-page or use the CMS to generate dynamic JSON-LD
+- Shopify: Product schema is auto-generated. Add Organization and Article manually.
+- Custom CMS: Generate JSON-LD server-side with a template that pulls real field values
+
+### Reference patterns
+See `references/implementation-patterns.md` for copy-paste JSON-LD for every schema type listed above.
+
+---
+
+## Common Mistakes
+
+These are the ones that actually matter — the errors that kill rich results eligibility:
+
+| Mistake | Why It Breaks | Fix |
+|---------|--------------|-----|
+| Missing `@context` | Schema won't parse | Always include `"@context": "https://schema.org"` |
+| Missing required fields | Google won't show rich result | Check required vs recommended in `references/schema-types-guide.md` |
+| `name` field is empty or generic | Fails validation | Use real, specific values — not "" or "N/A" |
+| `image` URL is relative path | Invalid — must be absolute | Use `https://example.com/image.jpg` not `/image.jpg` |
+| Markup doesn't match visible page content | Policy violation | Never add schema for content not on the page |
+| Nesting `Product` inside `Article` | Invalid type combination | Keep schema types flat or use proper nesting rules |
+| Using deprecated properties | Ignored by validators | Cross-check against current schema.org — types evolve |
+| Date in wrong format | Fails ISO 8601 check | Use `"2024-01-15"` or `"2024-01-15T10:30:00Z"` |
+
+---
+
+## Schema and AI Search
+
+This is increasingly the reason to care about schema — not just Google rich results.
+
+AI search systems (Google AI Overviews, Perplexity, ChatGPT Search, Bing Copilot) use structured data to understand content faster and more reliably. When your content has clean schema:
+
+- **AI systems parse your content type** — they know it's a HowTo vs an opinion piece vs a product listing
+- **FAQPage schema increases citation likelihood** — AI systems love structured Q&A they can pull directly
+- **Article schema with `author` and `datePublished`** — helps AI systems assess freshness and authority
+- **Organization schema with `sameAs` links** — connects your entity across the web, boosting entity recognition
+
+Practical actions for AI search visibility:
+1. Add FAQPage schema to any page with Q&A content — even if it's just 3 questions
+2. Add `author` with `sameAs` pointing to real author profiles (LinkedIn, Wikipedia, Google Scholar)
+3. Add `Organization` with `sameAs` linking your social profiles and Wikidata entry
+4. Keep `datePublished` and `dateModified` accurate — AI systems filter by freshness
+
+---
+
+## Testing & Validation
+
+Always test before publishing. Use all three:
+
+1. **Google Rich Results Test** — `https://search.google.com/test/rich-results`
+   - Tells you if Google can parse the schema
+   - Shows exactly which rich result types are eligible
+   - Shows warnings vs errors (errors = no rich result, warnings = may still work)
+
+2. **Schema.org Validator** — `https://validator.schema.org`
+   - Broader validation against the full schema.org spec
+   - Catches errors Google might miss or that affect other parsers
+   - Good for structured data targeting non-Google systems
+
+3. **`scripts/schema_validator.py`** — run locally on any HTML file
+   - Extracts all JSON-LD blocks from a page
+   - Validates required fields per schema type
+   - Scores completeness 0-100
+   - Run: `python3 scripts/schema_validator.py page.html`
+
+4. **Google Search Console** (after deployment)
+   - Enhancements section shows real-world errors at scale
+   - Takes 1-2 weeks to update after deployment
+   - The only place to see rich results performance data (impressions, clicks)
+
+---
+
+## Proactive Triggers
+
+Surface these without being asked:
+
+- **FAQPage schema missing from FAQ content** → any page with Q&A format and no FAQPage schema is leaving easy rich results on the table. Flag it and offer to generate.
+- **`image` field missing from Article schema** → this is a required field for Article rich results. Google won't show the article card without it.
+- **Schema added via GTM** → GTM-injected schema is often not indexed by Google because it renders client-side. Recommend server-side injection.
+- **`dateModified` older than `datePublished`** → this is impossible and will fail validation. Flag and fix.
+- **Multiple conflicting `@type` on same entity** → e.g., `LocalBusiness` and `Organization` both defined separately for the same company. Should be combined or one should extend the other.
+- **Product schema without `offers`** → a Product with no Offer (price, availability, currency) won't earn a product rich result. Flag the missing Offer block.
+
+---
+
+## Output Artifacts
+
+| When you ask for... | You get... |
+|---------------------|------------|
+| Schema audit | Audit report: schemas found, required fields present/missing, errors, completeness score per page, priority fixes |
+| Schema for a page type | Complete JSON-LD block(s), copy-paste ready, populated with placeholder values clearly marked |
+| Fix my schema errors | Corrected JSON-LD with change log explaining each fix |
+| AI search visibility review | Entity markup gap analysis + FAQPage + Organization `sameAs` recommendations |
+| Implementation plan | Page-by-page schema implementation matrix with CMS-specific instructions |
+
+---
+
+## Communication
+
+All output follows the structured communication standard:
+- **Bottom line first** — answer before explanation
+- **What + Why + How** — every finding has all three
+- **Actions have owners and deadlines** — no "we should consider"
+- **Confidence tagging** — 🟢 verified (test passed) / 🟡 medium (valid but untested) / 🔴 assumed (needs verification)
+
+---
+
+## Related Skills
+
+- **seo-audit**: For full technical and content SEO audit. Use seo-audit when the problem spans more than just structured data. NOT for schema-specific work — use schema-markup.
+- **site-architecture**: For URL structure, internal linking, and navigation. Use when architecture is the root cause of SEO problems, not schema.
+- **content-strategy**: For what content to create. Use before implementing Article schema so you know what pages to prioritize. NOT for the schema itself.
+- **programmatic-seo**: For sites with thousands of pages that need schema at scale. Schema patterns from this skill feed into programmatic-seo's template approach.