SERPsKit Amazon Autopilot Lab

The complete guide to your autopilot Amazon affiliate plugin — from installation to AI-generated articles with Claude Opus 4.7 and Article Profiles.

What this plugin does Turns any WordPress blog into a high-converting Amazon affiliate site. Fetches product data from Amazon, embeds polished product blocks in posts, and generates entire articles — both commercial roundups and informational guides — using Claude Opus 4.7 with adaptive thinking.
Amazon Autopilot Lab dashboard

Overview

🛒Product Library

Fetch Amazon products by ASIN or keyword. Auto-updates prices every 12–24 hours.

🎨5 Gutenberg Blocks

Product Box, Comparison Table, Product List, Pros & Cons, Deal Box.

🤖AI Autopilot

Keywords in → complete articles out. Uses Claude Opus 4.7 or OpenAI.

Article Profiles

Claude.ai Project-style profiles with persistent instructions and reference files for premium articles.

📊Click Tracking

Know which products and pages earn you commissions.

🔍SEO Schema

Automatic Product, Review, and AggregateRating markup.

What you need

  • WordPress 6.0+ running PHP 7.2 or higher
  • A license key from SERPsKit (format: SKA-XXXXX-XXXXX-XXXXX-XXXXX)
  • Amazon Associates account + PA-API credentials
  • An OpenAI or Anthropic Claude API key (required for AI features)

1. Installation

Upload the plugin

Plugins → Add New → Upload Plugin. Choose serpskit-autopilot-lab.zip and click Install Now.

Activate

A new menu called Autopilot Lab appears in your WordPress sidebar.

Verify

Go to Autopilot Lab → Dashboard. A "License Inactive" warning is expected — you'll fix it next.

Tip: The plugin auto-creates eight database tables on activation to store products, AI content, clicks, keywords, article profiles, and profile files.

2. License activation

Go to the License page

Autopilot Lab → License

Paste your license key

Format: SKA-XXXXX-XXXXX-XXXXX-XXXXX

Click "Activate License"

License activation process

License tiers

TierSitesBest for
Single1 websitePersonal blog
Five5 websitesSmall agency
Ten10 websitesMulti-site operator

Moving a license between sites

On the old site: Autopilot Lab → License → Deactivate on this site. Then activate on the new one.

3. Amazon PA-API setup

PA-API is how the plugin fetches live product data. Without it, only the scraper fallback works (risky).

Getting PA-API credentials

  1. Log into Amazon Associates
  2. Tools → Product Advertising API
  3. Manage Your Credentials → Add Credentials
  4. Copy Access Key and Secret Key (secret is shown once)
Requirement: Amazon grants PA-API access only to accounts with 3 qualifying sales in 180 days.

Entering credentials

Open Settings → Amazon PA-API tab

Autopilot Lab → Settings → 🔑 Amazon PA-API

Fill in three fields

  • Access Key ID
  • Secret Access Key
  • Partner Tag (usually your affiliate tag, e.g. yoursite-20)

Set Affiliate Tag in General tab

Save and test

Go to Autopilot Lab → Products, add a known ASIN like B08N5WRWNW.

Amazon Product Advertising PA-API settings

Supported Amazon locales

US, UK, CA, DE, FR, IT, ES, JP, IN, AU, BR, MX, AE.

Scraper fallback

Use at your own risk: Scraping violates Amazon TOS and may block your server IP. Only enable temporarily while waiting for PA-API access.

4. AI provider setup

Configure both OpenAI and Claude so one falls back to the other automatically.

Choosing a provider

ProviderStrengthsCost per article
Claude Opus 4.7 (recommended)Best writing quality, adaptive thinking~$0.05–0.30
Claude Sonnet 4.6Cheaper balance~$0.02–0.10
OpenAI GPT-4o-miniFast & cheap fallback~$0.01–0.02

Getting API keys

Both require billing credit before keys work ($5 minimum each).

Configuring

Open Settings → AI Providers

Autopilot Lab → Settings → 🤖 AI Providers

Pick primary and fallback

Recommended: Primary = Claude, Fallback = OpenAI.

Paste API keys and choose models

  • Claude model: claude-opus-4-7 (highest quality)
  • OpenAI model: gpt-4o-mini (best value fallback)

Enable Adaptive Thinking (recommended)

Claude automatically reasons longer on complex tasks. Set Effort to High.

Save settings

Claude Opus 4.7 model with Adaptive thinking mode
Custom model field: The "Custom model ID" input lets you paste any future Anthropic model (e.g. claude-opus-5-0) without waiting for a plugin update.

5. General & display settings

General tab

SettingDescription
Affiliate TagAppended to every Amazon link
Amazon LocaleWhich Amazon store to fetch from
Cache duration (hours)How long product data is cached (≤24h per Amazon TOS)

Display tab

SettingRecommended
Open links in new tab✓ On
Add rel="nofollow"✓ On
Add rel="sponsored"✓ On
Show affiliate disclosure✓ On (FTC requirement)

Dashboard

Autopilot Lab → Dashboard shows plugin health at a glance:

  • License status pill (top-right)
  • 4 stat cards — products cached, 7-day clicks, 30-day clicks, AI posts published
  • Top clicked products (30d)
  • Autopilot status — running or disabled, plus queue stats
  • Quick actions — shortcuts to Products, Autopilot, Article Profiles, Settings
Autopilot Lab main dashboard with products click stats

Product library

Autopilot Lab → Products holds all cached Amazon products. Products are added three ways:

  1. Manually via "Add Product" form
  2. Automatically when inserted into a block
  3. In bulk by Autopilot when searching for keywords

Adding a product manually

Find the ASIN

The 10-character code in any Amazon URL: amazon.com/dp/B08N5WRWNW/

Paste into the form

Select locale, click Fetch & Add.

Actions

  • ⟳ Refresh — re-fetches live data
  • ✕ Delete — removes from cache (re-fetches on next view)

Gutenberg blocks

All 5 blocks live in the Amazon Affiliate category. Open any post, click +, search "amazon".

All Amazon Gutenberg blocks

🛒 Product Box Single product

Single product card for inline mentions.

How to use

  1. Insert the Amazon Product Box block
  2. Enter ASIN in the placeholder or sidebar
  3. Block auto-fetches and previews
Amazon Product Box block in editor with sidebar settings

Key settings

SettingDescription
LayoutHorizontal or Vertical (card)
StyleModern, Minimal, Bold, Glass
CTA textButton label, e.g. "Check on Amazon"
BadgeRibbon text like "Editor's Choice"
TogglesShow/hide rating, price, features, Prime
Product Box block frontend render

📊 Comparison Table Multi-product · Sortable

The highest-converting block — compare 2–10 products side by side.

How to use

  1. Insert the block
  2. Add ASINs one by one (press Enter or click + Add)
  3. Remove any ASIN by clicking × next to its pill
Comparison Table block in editor with ASIN pills

Key settings

SettingDescription
Table titleOptional heading above the table
CTA textButton in the action column
Show rank columnAdds #1, #2, #3 badges
Sortable columnsUsers can click Rating/Price to sort
Highlight rowRow number for "Editor's Pick" marker
Comparison Table block frontend render with multiple products
Pro tip: Set Highlight to 1 to mark the first product as Editor's Pick — the best converting position.

📋 Product List Ranked roundup

Vertical ranked list for "Top 10" articles.

How to use

  1. Insert the block
  2. Add ASINs in order (first = #1)
  3. Use ↑ / ↓ to reorder

Key settings

  • List title — optional heading
  • Rank style: Numbered badge or Medal (🥇🥈🥉)
  • CTA text
Product List block frontend render

⚖️ Pros & Cons AI-filled · Multi-line

Two-column pros/cons grid. Fill manually or auto-generate from an ASIN.

Manual use

  1. Insert the block
  2. Type one pro per line — press Enter for new items
  3. Do the same in the Cons column

AI-powered use

  1. In the sidebar, enter an ASIN under Auto-fill with AI
  2. Click the purple ✨ Generate with AI button
  3. Wait 5–10 seconds — textareas fill automatically
  4. Edit as needed
Pros and Cons block frontend render
Cached: Once generated, pros/cons are cached per ASIN — reusing doesn't burn new AI tokens.

🔥 Deal Box Countdown · High urgency

Red-bannered deal card with optional live countdown.

How to use

  1. Insert the block
  2. Enter ASIN, deal title, CTA text
  3. For countdown: set expiry like 2026-12-31 23:59
Deal Box block frontend render with countdown timer
Don't fake scarcity: Only use countdowns for real deals.

Shortcodes (Classic Editor)

Single product

[sak_product asin="B08N5WRWNW" layout="horizontal" cta="Buy on Amazon" badge="Editor's Choice"]

Comparison table

[sak_comparison asins="B08N5WRWNW,B09XYZ1234,B07ABC5678"
                title="Best Headphones 2026"
                highlight="1"]

Product list

[sak_list asins="B08N5WRWNW,B09XYZ1234,B07ABC5678" title="Top 3 Picks"]

Deal box

[sak_deal asin="B08N5WRWNW" title="Black Friday Deal" expires="2026-11-29 23:59"]

Pros & cons

[sak_pros_cons asin="B08N5WRWNW"
               pros="Great sound|Long battery|Comfortable"
               cons="Pricey|Heavy"]

Pros and cons use pipe (|) as separator.

AI Article Generation — Two Workflows

The plugin offers two independent workflows for AI article generation. Use whichever fits the task:

🤖 Autopilot✨ Article Profiles
Best forFast bulk generationHighest-quality premium articles
InstructionsPer-batch textarea (non-persistent)Saved permanently, editable anytime
Reference files✓ TXT, MD, CSV, PDF, DOCX, JSON, XML, HTML
WorkflowQueue keywords → cron generatesClick "Generate Now" per article
ModelGlobal Claude settingGlobal OR per-profile override
Adaptive ThinkingGlobal settingPer-profile toggle
Prompt Caching✓ 90% cost savings
Scheduled runsManual only
Use both: Autopilot for volume; Article Profiles for flagship, high-intent articles that deserve maximum quality.

🤖 Autopilot — Fast Bulk Generation

Drops keywords into a queue. Cron picks them up and writes articles automatically.

Enable in Settings

Autopilot Lab → Settings → ✈️ Autopilot tab. Check Enable Autopilot.

Choose frequency

Hourly / Twice daily / Daily (recommended) / Weekly.

Posts per run

Start with 1 until you trust the output.

Post type & status

  • Post status: Draft (recommended) — review before publish

Default article type

  • 💰 Commercial — product roundup
  • 📖 Informational — general article
  • 🛠 How-to
  • 📋 Listicle
  • 📚 Guide

Global writing instructions (optional)

Applied to every AI-generated article. Per-keyword instructions override these.

Tone & length

Short (~800w), Medium (~1500w), Long (~2500w).

Adding keywords to the queue

Open Autopilot Lab → Autopilot

Paste keywords (one per line)

best wireless earbuds under 100
how to clean leather boots
top air fryers for small kitchens

Pick article type for this batch

Overrides the global default.

Category (optional)

Auto-created if it doesn't exist.

Custom writing instructions (optional)

Per-batch overrides for this keyword set.

Click "+ Add to Queue"

Autopilot page with keyword queue form and existing keywords table

Queue statuses

StatusMeaning
PendingWaiting for next run
ProcessingCurrently being written
CompletedPost created — click "View Post"
FailedError — check logs

Click ▶ Run next to any pending keyword to generate immediately.

The 7-step pipeline

  1. Amazon search — PA-API finds up to 8 matching products
  2. Product caching — each saved to your library
  3. AI article generation — Claude/OpenAI writes title, intro, sections, FAQ, conclusion
  4. Pros/cons generation — each product gets AI pros/cons (commercial only)
  5. Block assembly — post is built with all blocks embedded
  6. Featured image — downloaded from first product
  7. Publication — saved as draft or published

Typical time: 20–90 seconds per article (longer with thinking enabled).

Quality tips

Critical: Always keep posts as Draft and review before publishing. Google penalizes unreviewed AI content.

Better keywords

❌ Too broad✅ Specific
headphonesbest noise-cancelling headphones under 200
knivesbest chef knives for home cooks 2026

Review checklist before publishing

  • Does the intro hook the reader?
  • Are pros/cons realistic or generic?
  • Do descriptions mention specific features?
  • Are prices still accurate?
  • Replace blurry featured images
  • Add your own personal experience paragraph — biggest ranking lift
  • Fix AI clichés ("In the world of...", "When it comes to...")

✨ Article Profiles — Premium Quality

What is this? Article Profiles replicate Claude.ai's Projects feature inside WordPress. Each profile has persistent instructions and reference files (sitemap, style guide, etc.) that get sent with every article generation — the same as chatting with a Claude.ai Project. Ideal for your most important articles.

When to use Profiles vs Autopilot

  • Use Autopilot when: speed matters, many keywords, general-purpose articles
  • Use Profiles when: quality matters, consistent voice/brand, reference material needed (sitemap for internal linking, style guide, brand kit)

Create your first profile

Go to Article Profiles

Autopilot Lab → ✨ Article Profiles

Click "+ New Profile"

Name it

Example: "My Tech Blog" — for your reference only.

Save basic defaults

  • Default tone, length, article type
  • Optional Claude model override (e.g. claude-opus-4-7)
  • Enable Adaptive Thinking ✓
  • Enable Prompt Caching ✓
New Profile form filled in name, description, tone, length

Persistent Instructions

The most important field. These instructions are sent with every article generation request for this profile — exactly like the Instructions box in a Claude.ai Project.

What to put here

  • Voice and persona: "You are an experienced [topic] expert writing for yourdomain.com..."
  • Style rules: first-person vs third-person, tone, sentence length
  • Brand details: name, URL, tagline, anything to reference
  • Formatting rules: heading style, paragraph length, list formatting
  • Internal linking guidance: "Always check sitemap.txt in this profile and link to relevant existing articles where natural"
  • Words/phrases to avoid: AI clichés, over-used terms
  • SEO rules: keyword density, meta description format
Copy from Claude.ai: If you already have a working Claude.ai Project, copy its Instructions content directly into this field.

Reference Files

Upload files that support article writing. Each file's text is extracted and sent as context with every article.

Supported file types

TypeBest for
.txt, .mdSitemap, link lists, simple style guides
.csvData tables, product lists
.docxContent strategy docs, brand guidelines
.pdfPDF guides (text-based only)
.json, .xmlStructured data
.htmlReference web pages (stripped to text)
PDF limitation: Text extraction only works on text-based PDFs. Scanned/image PDFs will fail — upload a .txt or .docx version instead.

Typical files for a content site

  1. sitemap.txt — list of all existing article URLs (for internal linking)
  2. style_guide.docx — voice, tone, formatting rules
  3. brand_kit.txt — brand name, tagline, key phrases
  4. semantic_content_writing_strategy.docx — SEO strategy

Upload a file

In the profile edit page, scroll to "Reference Files"

Click "Choose file"

Click "+ Upload File"

Plugin extracts text and shows token count.

Reference Files section with upload form and file list table

Remove a file

Click the red ✕ button next to any file in the list.

Limits

  • 5 MB per file
  • ~200K total tokens per profile (safety cap)

Generate an article with a profile

Scroll to "Generate Article With This Profile"

Orange-highlighted section at the bottom of the profile page.

Enter keyword / topic

Example: how to start growing sweet corn in spring

Optional: override article type

Leave as "Use profile default" or pick a different type for this one article.

Optional: category

Click "✨ Generate Now"

Takes 30–90 seconds with thinking enabled.

Success notice appears

Click "Edit post →" to review the draft.

Cost & caching

The profile page shows live cost estimates based on your attached files:

  • Total tokens — sum of all file tokens
  • First article cost — full input cost (Opus 4.7: $5/M tokens)
  • Cached article cost — 10% of full price after prompt cache hit

How prompt caching works

With Prompt Caching enabled, the first article you generate in a 5-minute window pays full price for the file context. Subsequent articles within that window pay 10% for the same context. Big savings when generating multiple articles back-to-back.

Example costs (Opus 4.7)

Files attachedTokensFirst articleCached
Just sitemap.txt~3,000~$0.015~$0.0015
Sitemap + style guide~8,000~$0.04~$0.004
Full brand bundle~25,000~$0.125~$0.0125

Add output tokens (roughly 4,000 tokens per article = $0.10) to get total cost.

Tip: Generate 5–10 articles in one session to maximize cache hits and minimize cost per article.

Click tracking

Every affiliate link is wrapped with a tracking redirect: yoursite.com/?sak_go=ASIN&sak_loc=US&sak_pid=123

When clicked, plugin logs ASIN, post ID, device, referrer, hashed IP → redirects to Amazon with your tag.

Viewing stats

  • Dashboard — total clicks (7d / 30d), top 5 products
  • Top clicked products table — click count per ASIN
Privacy: IPs are SHA-256 hashed with your AUTH_KEY. Raw IPs never stored. GDPR-friendly.

SEO & Schema.org markup

JSON-LD structured data is injected automatically for any post with Amazon blocks. Google uses this for rich snippets (stars, prices, images).

  • Single product: full Product schema with AggregateRating, Offer, Brand
  • Multiple products: ItemList wrapping all products

Testing markup

Use the Google Rich Results Test — should show "Product snippet detected".

Cron & automatic updates

EventFrequencyPurpose
sak_price_updateTwice dailyRefreshes up to 20 stale products
sak_autopilot_runHourlyProcesses pending keywords
sak_process_queueEvery 5 minProcesses the autopilot task queue

Real cron (for low-traffic sites)

WP-Cron only runs when someone visits. Add a real cron job:

*/5 * * * * wget -q -O - https://yoursite.com/wp-cron.php?doing_wp_cron >/dev/null 2>&1

Then in wp-config.php:

define('DISABLE_WP_CRON', true);

REST API endpoints

For developers — all under /wp-json/sak/v1/, require WP nonce + role.

MethodEndpointPurpose
GET/product/{asin}Get product data
GET/search?keywords=…Amazon keyword search
POST/ai/pros-consGenerate pros/cons for ASIN
POST/ai/summaryGenerate product summary
POST/autopilot/generateTrigger immediate generation
GET/autopilot/keywordsList keyword queue
POST/autopilot/keywordsAdd bulk keywords
GET/statsDashboard statistics

Troubleshooting

License won't activate

  • Verify key typed exactly (no spaces)
  • Check server clock is correct (HMAC is time-aware)
  • Confirm activation limit not reached

"No products found" for every ASIN

  • Re-check PA-API credentials
  • Confirm Partner Tag is approved for PA-API
  • Need 3+ qualifying sales in 180 days
  • Try known ASIN B08N5WRWNW

AI generation fails

  • Test API key on provider's console
  • Check billing has credit loaded
  • Try fallback provider
  • For JSON errors: switch to claude-opus-4-7 or gpt-4o

Article Profile upload fails

  • Check file is under 5 MB
  • Confirm file extension is in allowed list
  • For PDF errors: try .txt or .docx version
  • DOCX: requires PHP ZipArchive extension (standard on most hosts)

Profile generation very slow

With Adaptive Thinking enabled, Opus 4.7 can take 60-120 seconds. Disable thinking for faster (but lower quality) output, or use a cheaper model in Model Override.

Gutenberg blocks don't appear

  • Clear browser cache (Ctrl+Shift+R)
  • Disable other plugins to find conflicts
  • Check browser console for JS errors

Autopilot generates empty posts

  • Keyword too narrow — broaden it
  • AI returned invalid JSON — try claude-opus-4-7 or gpt-4o

FAQ

What's the difference between Autopilot and Article Profiles?

Autopilot is for fast bulk generation (queue many keywords, cron writes them). Article Profiles is for premium quality — you create a profile with persistent instructions and reference files, then generate articles one at a time with maximum context. See the comparison table.

Can I use this without PA-API?

Yes via scraper fallback, but violates Amazon TOS and isn't recommended.

Will AI-generated content rank on Google?

Yes IF you review and add human value. Google's Helpful Content guideline isn't "no AI" — it's "helpful and original". Always add personal experience before publishing.

How much do Article Profiles cost vs regular Autopilot?

Profiles with Opus 4.7 + reference files cost more per article (~$0.10-0.30) vs Autopilot with cheaper models (~$0.01-0.05). Worth it for flagship articles where quality matters most.

Can I import my Claude.ai Project?

Not directly — Anthropic's API doesn't expose Projects. But you can copy the Instructions text and re-upload the same files into an Article Profile. The functionality is equivalent.

Does Prompt Caching really save 90%?

Yes, for cached tokens. First article in a 5-minute window pays full price for file context; subsequent articles pay 10%. Output tokens (the article itself) aren't cached.

Can I use Claude Opus 5 or newer when released?

Yes. Use the "Custom model ID" field in Settings to paste any Anthropic model ID without waiting for a plugin update.

Can I move my license between sites?

Yes. Deactivate on old site first, then activate on new.

What happens if my license expires?

Plugin continues working but license status turns "invalid" and REST endpoints stop. Existing posts still render.

How do I customize block designs?

All CSS uses variables. Add to theme's Additional CSS:

:root {
  --sak-primary: #e91e63;
  --sak-primary-dark: #c2185b;
}

Will Amazon ban my account?

Not if you use PA-API (not scraping), refresh prices within 24 hours (automatic), disclose affiliate relationships (plugin does this), and don't make false claims.

How do I backup my data?

Standard WordPress backups cover everything. All plugin data is in tables prefixed wp_sak_.

SERPsKit Amazon Autopilot Lab — User Guide v1.3.0 — Visit serpskit.com