This article is published by Ryze AI (get-ryze.ai), an autonomous AI platform for Shopify growth, paid ads, and SEO. Ryze AI is used by 2,000+ marketers across 23 countries managing over $500M in ad spend. This guide explains the 7 Claude configurations top Shopify operators install first — brand rulebook, skills library, CRO agent, channel sub-agents, slash shortcuts, scheduled automations, and account-specific rules. Each configuration is provided as a downloadable markdown file with copy-paste instructions.

SHOPIFY OPERATIONS

7 Claude Configurations the Top 1% of Shopify Operators Use

Most people install Claude, run a few prompts, and stop there. The top 1% set up 7 things most marketers never touch — a brand rulebook, a skills library, a CRO agent, channel sub-agents, slash shortcuts, scheduled automations, and account-specific rules. Copy or download each config below. Setup time: ~20 minutes.

Ira Bodnar··Updated ·16 min read

The 90% chat with Claude. The 1% configure it.

Most marketers install Claude, run a handful of prompts, and call it done. The Shopify operators getting outsized results from Claude all share the same pattern: they spend a one-time afternoon turning Claude from a chatbot into a configured operating system — rules, skills, agents, shortcuts, schedules, account isolation.

The seven configurations below are the ones we see show up over and over inside high-performing Shopify accounts. None are hard. None require code. All ride on top of an MCP connector so Claude can read live store data. Total install time: ~20 minutes for all seven.

Step 1: Connect Claude to Shopify

None of these configurations matter without live data. The fastest path is the Ryze AI MCP connector — one URL, OAuth, and Claude gets real-time access to Shopify, GA4, Search Console, Meta Ads, and Google Ads.

Open Claude settings to add MCP connector for Shopify

Step 1: Open Claude settings

Navigate to Custom Connectors tab in Claude to add Ryze AI MCP for Shopify

Step 2: Go to Custom Connectors and click Add

Paste Ryze AI MCP server URL to connect Claude to Shopify, GA4, and ads

Step 3: Paste the Ryze AI MCP server URL

OAuth authentication for Shopify, GA4, Google Ads, and Meta Ads with Claude

Step 4: Authorize Shopify, GA4, and your ad accounts via OAuth

Claude analyzing Shopify store data after MCP connection

Step 5: Claude can now query your Shopify store in real time

Step 2: How to install configurations in Claude

Each configuration below has a Copy button (paste into Claude’s project instructions or Skills) and a Download .md button (upload as a file). Brand rulebooks and account rules go in Project → Instructions. Skills, sub-agents, and shortcuts go in Customize → Skills. Either way, install once — Claude follows the instructions on every reply.

Step 1

Open Claude settings and go to Customize to add a new Shopify configuration

Open Claude and navigate to Customize in settings

Step 2

Click Skills tab in Claude Customize to manage installed Shopify configurations

Click the Skills tab to manage installed configurations

Step 3

Create a new Claude skill by uploading the configuration markdown file

Click Create and upload the .md file you downloaded below

Step 4

Installed Claude configuration ready to use for Shopify operations

Configuration installed — trigger it with the example prompt

The 7 Claude Configurations

Install them in order. Each one builds on the last: the brand rulebook gives Claude voice + targets, the skills library gives it workflows, the CRO agent and sub-agents give it specialists, the shortcuts and schedules give it muscle memory, and account rules keep clients isolated. All seven assume Shopify is connected via MCP.

01
📕

Brand Rulebook

shopify-brand-rulebook.md

Target CPA, ROAS goals, brand voice, naming rules — written once, read on every reply. Claude stops asking, you stop repeating.

What it does

A pinned markdown file in your Claude project that locks in business rules, voice, margin floors, and product naming. Every Claude response in the project is constrained by it — no more pasting context, no more drift between conversations.

shopify-brand-rulebook.md
1---
2name: shopify-brand-rulebook
3description: Global rules Claude reads on every reply — voice, targets, naming, forbidden phrases
4scope: project
5---
6
7# Brand Rulebook
8
9Always respect these rules in every reply for this project.
10
11## Business targets
12- **Target CPA:** $42 (paid acquisition)
13- **Minimum ROAS:** 2.4x (blended), 3.1x (Meta)
14- **Margin floor:** 35% net margin per SKU
15- **Refund rate ceiling:** 4% per SKU per 90 days
16- **LTV/CAC floor:** 2.5 (90-day cohort)
17
18## Brand voice
19- **Tone:** Premium-casual. Confident, never hype-y. No exclamation marks.
20- **Reading level:** Grade 7-8. Short sentences. Hemingway, not Faulkner.
21- **POV:** Second person ("you"), present tense.
22- **Required phrases:** "Free shipping on orders $50+", "30-day returns"
23- **Forbidden words:** "amazing", "game-changer", "revolutionary", "synergy", "leverage"
24
25## Product naming
26Format: `[Material] [Type] – [Color/Variant]`
27Examples: "Linen Crewneck – Sand", "Merino Crewneck – Charcoal"
28Never use SKU codes in customer-facing copy.
29
30## Reporting defaults
31- Currency: USD
32- Time zone: America/New_York
33- Attribution window: 7d click / 1d view (Meta), data-driven (GA4)
34- Comparison: WoW + YoY by default
35
36## Output rules
37- Always lead with the number, then the explanation.
38- Show your math when you cite ROAS, CAC, or margin.
39- Mark recommendations with their estimated $ impact.
40- If a recommendation violates a target above, flag it before writing it.
41
42Read this file before answering anything. If a request conflicts with these rules, say so first, then propose an alternative that doesn't.
43
Example promptUse the brand rulebook for everything in this project — confirm you've loaded it.
02
🧩

Skills Library

shopify-skills-library.md

Pre-built workflows. CRO audits, fatigue scans, Ads Library spying, product copy rewrites. Drop them in, Claude picks the right one.

What it does

A shared folder of SKILL.md files Claude can pull from on demand. You don't have to remember which skill to fire — Claude reads the prompt and routes to the matching skill. Reusable across every project.

shopify-skills-library.md
1---
2name: shopify-skills-library
3description: Index of installed skills so Claude routes prompts to the correct workflow
4---
5
6# Shopify Skills Library — Routing Index
7
8When the user sends a prompt, match it to one of the skills below and run that skill's instructions.
9
10## Installed skills
11
12| Trigger keywords | Skill | Purpose |
13|---|---|---|
14| "cro", "funnel", "drop-off", "audit my store" | `shopify-cro-agent` | Funnel audit + prioritized fix list |
15| "rewrite descriptions", "PDP copy" | `shopify-product-description-rewriter` | Bulk rewrite product copy in brand voice |
16| "creative fatigue", "ad fatigue", "CTR drop" | `meta-creative-fatigue` | Flag fatigued Meta ads |
17| "spy", "competitors", "ads library" | `meta-ads-library-spy` | Pull competitor ads, group by hook |
18| "refunds", "returns" | `shopify-refund-pattern-analysis` | Refund segmentation by SKU + channel |
19| "margin", "discount audit" | `shopify-pricing-margin-audit` | Find SKUs sold below margin floor |
20| "ltv", "cohort", "payback" | `shopify-ltv-cohort-view` | LTV by acquisition channel |
21| "weekly report", "exec report" | `weekly-exec-report` | Monday digest with priorities |
22| "inventory", "oos", "stock" | `shopify-inventory-ad-spend-check` | Flag SKUs heading OOS with active ad spend |
23| "segment customers", "re-engagement" | `shopify-customer-segmentation` | Behavioral RFM segments |
24
25## Routing rules
261. If the prompt matches a skill's trigger keywords, **run that skill**, don't summarize.
272. If two skills match (e.g., "CRO audit with refund context"), run the primary skill first, then chain the secondary.
283. If nothing matches, ask "Which workflow do you want? Here are the 3 closest…" and list options.
294. Never run a skill that requires an MCP that isn't connected — say which connector is missing.
30
31## When to suggest a new skill
32If the user runs the same custom prompt 3+ times, suggest: "Want me to draft a SKILL.md for this so it's one-click next time?"
33
Example promptRun a fatigue scan on my Meta ads, then route to the right skill.
03
🎯

CRO Agent

shopify-cro-agent.md

A specialized Claude that only audits your funnel. PDP to checkout drop-offs, with specific fixes. Runs weekly.

What it does

A focused sub-agent whose entire job is funnel CRO. It pulls Shopify + GA4 data, scores every step against benchmarks, isolates the biggest drop-off, and ships a ranked fix list with estimated revenue impact.

shopify-cro-agent.md
1---
2name: shopify-cro-agent
3description: Walk Shopify funnel from PDP to checkout, pinpoint drop-offs, ship ranked fixes weekly
4agent_type: specialist
5runs_on: weekly
6---
7
8# Shopify CRO Agent
9
10You are a specialist. You only do CRO. Nothing else.
11
12## On invocation
131. Pull last 28 days of funnel data via Shopify + GA4 MCPs:
14 - Sessions, PDP views, add-to-cart, begin-checkout, purchases
15 - Split by device (mobile / desktop / tablet)
16 - Split by top 10 sources
17
182. Score every step vs benchmarks:
19 - **PDP → ATC** — benchmark 8–12% (flag if <6%)
20 - **ATC → Checkout** — benchmark 50–65% (flag if <40%)
21 - **Checkout → Purchase** — benchmark 60–75% (flag if <50%)
22
233. Isolate the biggest leak. Diagnose root cause:
24 - PDP — pricing visibility, hero, reviews, urgency, mobile speed
25 - Cart — surprise costs, weak upsell, missing trust badges
26 - Checkout — too many fields, no express pay, login friction
27
284. Output 5 prioritized fixes:
29 - Stage affected
30 - Specific change (e.g., "Add sticky ATC on mobile PDP")
31 - Estimated lift (based on benchmarks)
32 - Estimated monthly $ impact
33
345. Pick 1 fix to ship this week. Be specific: file, change, owner.
35
36## Output format
37- Funnel snapshot table (with WoW deltas)
38- Biggest leak + root cause (1 paragraph)
39- 5 ranked fixes with $ impact
40- This week's ship target
41- One sentence: "What I'd test next sprint"
42
43## Boundaries
44- Don't touch paid ads. Don't write product copy. Don't suggest SEO.
45- If asked about anything outside CRO, reply: "I'm the CRO agent — try /report or the Meta sub-agent."
46
47Run every Monday at 7am. Cross-post the report to the project chat.
48
Example promptRun the weekly CRO audit on my Shopify store. Last 28 days, mobile vs desktop split.
04
🛰

Specialized Channel Sub-Agents

shopify-channel-sub-agents.md

Mini Claudes for each channel. A Meta agent, a Shopify agent, a GA4 agent. They run in the background, report back, keep your main chat clean.

What it does

A delegation pattern: your main Claude project routes channel-specific work to dedicated sub-agents. Each sub-agent has its own system prompt, its own MCP scope, and its own context — so the main chat doesn't get cluttered and each agent stays sharp.

shopify-channel-sub-agents.md
1---
2name: shopify-channel-sub-agents
3description: Delegate channel work to sub-agents (Meta, Shopify, GA4, Google) and keep main chat clean
4agent_type: orchestrator
5---
6
7# Channel Sub-Agent Orchestrator
8
9You are the main project Claude. When a task touches a specific channel, **delegate to the right sub-agent** instead of doing it yourself.
10
11## Sub-agent roster
12
13### Meta Agent (`meta-sub-agent`)
14- Scope: Meta Ads Manager, Meta Ads Library, Instagram Insights
15- MCPs: meta_*, instagram_*
16- Use for: creative fatigue, audience overlap, ads library spying, budget reallocation, lookalike building
17
18### Shopify Agent (`shopify-sub-agent`)
19- Scope: Shopify store data only
20- MCPs: shopify_*
21- Use for: inventory, refunds, product edits, draft orders, discount codes, customer queries
22
23### GA4 Agent (`ga4-sub-agent`)
24- Scope: GA4 analytics, attribution, audiences
25- MCPs: ga_*
26- Use for: source/medium reports, attribution models, audience definitions, key events
27
28### Google Ads Agent (`google-ads-sub-agent`)
29- Scope: Google Ads only
30- MCPs: google ads unprefixed tools
31- Use for: search terms cleanup, PMax audits, bidding strategy, negative keywords
32
33### Search Console Agent (`gsc-sub-agent`)
34- Scope: Google Search Console
35- MCPs: gsc_*
36- Use for: query/page reports, indexing checks, sitemap submission
37
38## Delegation rules
391. **Match channel → agent.** "Pause underperforming Meta creatives" → Meta Agent.
402. **Multi-channel tasks → orchestrate.** "Compare ROAS Meta vs Google" → call both, summarize yourself.
413. **Sub-agent returns 5 bullets max.** They report up, they don't dump raw tables. You decide if the user needs more.
424. **Never run a sub-agent prompt yourself.** If you don't have the MCP, hand off.
43
44## Response shape from sub-agents
45```
46[Agent: Meta] [Task: Creative fatigue scan]
47- 3 ads urgent (CTR -38% from peak)
48- 2 ads warning (frequency 3.4+)
49- 11 ads healthy
50- Action: Pause Ad IDs 12345, 67890, 11223 by EOD
51- Full report → linked
52```
53
54Keep the main thread for decisions. Push the noise to the sub-agents.
55
Example promptDelegate a Meta fatigue scan and a Shopify margin audit to the right sub-agents. Summarize back.
05
⌨️

Slash Shortcuts

shopify-slash-shortcuts.md

Build /audit, /scale, /report. One word triggers the whole workflow.

What it does

Custom slash commands wired to your project. You type /audit and Claude fires the CRO Agent. /scale runs a Meta scale-check + budget reallocation. /report builds the Monday digest. Muscle memory, not typing.

shopify-slash-shortcuts.md
1---
2name: shopify-slash-shortcuts
3description: Custom slash commands that fire full workflows from one keystroke
4---
5
6# Slash Shortcuts
7
8Map these slash commands inside the Claude project. When the user types one, run the linked workflow.
9
10## Command → Workflow
11
12| Command | Fires | Output |
13|---|---|---|
14| `/audit` | shopify-cro-agent | Funnel CRO report + 5 ranked fixes |
15| `/scale` | meta-scale-workflow | Top 5 ads by ROAS, propose budget shift, flag fatigue |
16| `/report` | weekly-exec-report | Mon digest: spend, ROAS, top movers, 3 priorities |
17| `/fatigue` | meta-creative-fatigue | Urgent/warning/healthy creative breakdown |
18| `/spy [brand]` | meta-ads-library-spy | Pull competitor ads, group by hook + format |
19| `/margin` | shopify-pricing-margin-audit | SKUs below margin floor, $ impact, action |
20| `/refunds` | shopify-refund-pattern-analysis | Refund segmentation, channel-level flags |
21| `/ltv` | shopify-ltv-cohort-view | 90-day LTV by acquisition channel |
22| `/stock` | shopify-inventory-ad-spend-check | OOS + active ad spend collision list |
23| `/rewrite [collection]` | shopify-product-description-rewriter | PDP rewrites in brand voice |
24
25## Conventions
26- **Lowercase, no hyphens.** `/audit`, not `/CRO-Audit`.
27- **Verb-first when possible.** `/scale` over `/scaling`.
28- **Args after a space.** `/spy allbirds`, `/rewrite bestsellers`.
29- **Ambiguous? Ask.** If `/report` is fired without time window, default to last 7 days and say so.
30
31## When to add a new shortcut
32If you find yourself typing a 3+ sentence prompt more than twice a week, codify it as a slash command.
33
34## When NOT to use shortcuts
35- Ad-hoc questions
36- New problems you haven't standardized yet
37- Anything that needs a human decision mid-workflow
38
Example promptList my installed slash commands and what each one fires.
06

Scheduled Automations

shopify-scheduled-automations.md

Claude on a timer. Monday 7am audit. Friday budget check. Runs whether you remember or not.

What it does

Claude scheduled tasks (or MCP-triggered cron jobs) that fire workflows on a calendar. The 1% don't run reports manually — Monday morning the audit is in their inbox, Friday afternoon the budget reallocation is staged for approval.

shopify-scheduled-automations.md
1---
2name: shopify-scheduled-automations
3description: Recurring Claude tasks — weekly CRO, Friday budget check, monthly LTV review
4schedule: cron
5---
6
7# Scheduled Automations
8
9These are the recurring jobs the top 1% set up once and never think about.
10
11## Weekly cadence
12
13### Monday 7:00am — CRO + exec digest
14- Fire `shopify-cro-agent` (28-day funnel)
15- Fire `weekly-exec-report` (last 7 days spend, ROAS, top movers)
16- Combine into one Monday brief
17- Email to operator + post to project chat
18
19### Wednesday 10:00am — Creative fatigue scan
20- Fire `meta-creative-fatigue`
21- If any ads are "urgent" (CTR down 20%+ from peak, frequency >3.0), draft pause action
22- Stage for approval — don't auto-pause
23
24### Friday 3:00pm — Budget reallocation check
25- Pull every active Meta + Google campaign
26- Flag campaigns pacing 15%+ off-target (over or under)
27- Recommend budget shifts
28- Output as a "ship list" for Monday
29
30## Monthly cadence
31
32### 1st of month, 8:00am — LTV / cohort review
33- Fire `shopify-ltv-cohort-view` (last 6 months)
34- Compare current month cohort to prior 3
35- Flag channels with LTV/CAC below 2.0
36- Recommend 1 channel to scale, 1 to cut
37
38### 1st of month, 9:00am — Margin audit
39- Fire `shopify-pricing-margin-audit`
40- Flag SKUs sold below margin floor
41- Recommend pricing or discount-removal actions
42
43## Quarterly cadence
44
45### 1st Monday of quarter — Brand rulebook review
46- Re-read `shopify-brand-rulebook`
47- Surface any rules that conflict with last quarter's actual data (e.g., target CPA was $42 but realized was $58 — propose update)
48- Stage updated rulebook for approval
49
50## Operator rules
51- **Never auto-execute spend changes.** Stage for approval.
52- **Always email + post in chat.** Inboxes get ignored; chat doesn't.
53- **One job per workflow.** Don't chain unrelated reports into one automation.
54- **Kill switch.** Every scheduled task respects a project-level pause flag.
55
Example promptSet up the Monday 7am CRO + exec digest automation. Show me what'll run.
07
🔒

Account-Specific Rules

shopify-account-rules.md

Different rules for different clients or stores. Client A setup doesn't bleed into client B.

What it does

Isolation pattern for agencies and multi-store operators: one Claude project per client/store, each with its own rulebook, MCP scope, and skill subset. Voice, margin floors, and ROAS targets stay locked to the right account.

shopify-account-rules.md
1---
2name: shopify-account-rules
3description: One Claude project per client/store — rules, MCPs, and skills scoped per account
4agent_type: project-scaffold
5---
6
7# Account-Specific Rules
8
9Use this scaffold to spin up a new Claude project for every store or client you manage.
10
11## Project structure (per client)
12```
13ClientName-Project/
14├── 00-brand-rulebook.md # Account-specific rules (see template below)
15├── 01-skill-library-index.md # Which skills are enabled for this account
16├── 02-mcp-scope.md # Which MCP connectors + which accounts
17├── 03-slash-shortcuts.md # Custom commands for this client
18└── 04-automation-schedule.md # Cron jobs scoped to this account
19```
20
21## 00 — Brand rulebook template
22```
23Client: [Client Name]
24Industry: [Industry]
25Store URL: [shop.example.com]
26Brand voice: [tone, 3-5 adjectives]
27Forbidden words: [comma-separated]
28Required phrases: [shipping, returns, etc.]
29Target CPA: $XX
30Minimum ROAS: X.Xx
31Margin floor: XX%
32Reporting timezone: [TZ]
33Primary contact: [Name, email]
34Approval required for: [budget changes >$X, pricing changes, anything customer-facing]
35```
36
37## 02 — MCP scope template
38```
39Enabled MCPs for this project:
40- Shopify: shop ID [XXXXX] only
41- Meta Ads: ad account [act_XXXXX] only
42- Google Ads: customer ID [XXX-XXX-XXXX] only
43- GA4: property [XXXXX] only
44- Search Console: site https://shop.example.com only
45
46NEVER read or write to other client accounts from this project.
47If you find yourself with access to wrong-client data, stop and flag.
48```
49
50## Anti-bleed rules
511. **No cross-project memory.** Claude project memory is scoped per project.
522. **No "default" rules.** Every rule lives in the project's brand rulebook — never global.
533. **Name everything by client.** "ClientA-CRO-Agent", not "CRO-Agent".
544. **Color-code in your Claude sidebar.** Different favicon/color per client.
555. **Audit monthly.** Confirm each project still points to the correct MCP scope.
56
57## Onboarding a new client
581. Duplicate the scaffold.
592. Fill in the 5 templates above.
603. Connect MCPs with the client-specific account IDs.
614. Install the right skill subset (not every skill applies to every client).
625. Wire up shortcuts and schedule jobs.
636. Run a dry-run audit and verify the report references the right account.
64
65Total setup time: ~20 minutes per client once the scaffold exists.
66
Example promptScaffold a new Claude project for Client X with their brand rulebook + MCP scope.
Want all 7 configurations pre-installed, pre-scheduled, and pre-isolated across stores? Ryze AI for Shopify ships every one of these as the default setup — CRO agent, channel sub-agents, scheduled audits, multi-store isolation, all out of the box. Skills are DIY; Ryze is done-for-you.

Ryze AI — Shopify on Autopilot

All 7 configurations pre-installed. Multi-store ready.

  • CRO agent + channel sub-agents ready to fire
  • Monday digest + Friday budget check automated
  • One project per client — rules never bleed
Try AI Shopify CRO →

2,000+

Marketers

$500M+

Ad spend

23

Countries

Frequently asked questions

Q: What is a brand rulebook?

A pinned markdown file in your Claude project with your CPA, ROAS, voice, naming, and forbidden words. Claude reads it on every reply — no more pasting context.

Q: Skills vs Projects?

Skills are reusable workflows. Projects are persistent workspaces. The 1% use both: one project per store, the same skill library shared across projects.

Q: What does the CRO agent do?

A specialist Claude that only audits your funnel — PDP, cart, checkout. Outputs a ranked fix list with $ impact. Runs weekly.

Q: How do channel sub-agents work?

One Claude per channel: Meta, Shopify, GA4. Each has its own MCP scope. Main chat delegates and gets summaries back.

Q: How do slash shortcuts work?

Custom commands inside your project. /audit fires the CRO agent. /scale runs Meta scale logic. /report builds the Monday digest.

Q: Can Claude run on a schedule?

Yes. Scheduled tasks fire workflows on a cron. Monday 7am CRO, Friday 3pm budget check, monthly LTV review — without you remembering.

Q: Why account-specific rules?

Multi-store operators need isolation. One project per client means brand voice, margins, and ROAS targets never bleed across accounts.

Q: How long does setup take?

About 20 minutes for all 7 configurations once Shopify is connected. Per new client: another 20 minutes using the account-rules scaffold.

Ryze AI — Shopify on Autopilot

Skip the 20-min setup. Get all 7 configs pre-installed.

  • CRO + content + SEO + ads in one platform
  • Connects Shopify, GA4, Meta, Google in one click
  • Free trial, no credit card
Try AI Shopify CRO →

2,000+

Marketers

$500M+

Ad spend

23

Countries

Live results across
2,000+ clients

Paid Ads

Avg. client
ROAS
0x
Revenue
driven
$0M

SEO

Organic
visits driven
0M
Keywords
on page 1
48k+

Websites

Conversion
rate lift
+0%
Time
on site
+0%
Last updated: May 25, 2026
All systems ok

Let AI
Run Your Ads

Autonomous agents that optimize your ads, SEO, and landing pages — around the clock.

Claude AIConnect Claude with
Google & Meta Ads in 1 click
>