Install
$ agentstack add skill-storehand-storehand-storehand-setup ✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ● Network access Used
- ● Filesystem access Used
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
Verified badge
Passed review? Show it. Paste this badge into your README, it links to the public security report.
Reliability & compatibility
Declared compatibility
Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.
We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.
How agent discovery & health will work →About
StoreHand setup
Get the user from a fresh install to a working, connected store profile. Aim for under ten minutes.
Network: the localhost auth callback (route 2 in shared/safety.md)
Two kinds of path, do not mix them up. This plugin's own files — shared/safety.md and the templates/ directory — live under the plugin's install directory. The profile you are about to create (.storehand/) goes in the user's working directory.
Step 0 — Find the plugin
$CLAUDE_PLUGIN_ROOT is empty in the environment of a Bash tool call. That is measured, not assumed; the working routes and the reasoning are in shared/plugin-root.md. Run this once, before anything else:
ROOT=$( {
printf '%s\n' "${CLAUDE_PLUGIN_ROOT:-}"
printf '%s' "$PATH" | tr ':' '\n' | sed -n 's|/bin$||p' | grep -E '/storehand/[^/]+$'
node -e 'const fs=require("fs"),os=require("os"),p=require("path");try{const j=JSON.parse(fs.readFileSync(p.join(os.homedir(),".claude","plugins","installed_plugins.json"),"utf8"));for(const[k,v]of Object.entries(j.plugins||{}))if(k.split("@")[0]==="storehand"&&v[0]&&v[0].installPath){console.log(v[0].installPath);break}}catch{}' 2>/dev/null
} | while IFS= read -r c; do
[ -n "$c" ] && [ -f "$c/shared/api-version.md" ] && { printf '%s' "$c"; break; }
done )
[ -n "$ROOT" ] && echo "StoreHand plugin root: $ROOT" || echo "StoreHand plugin root NOT FOUND"
Shell state does not survive between tool calls, so there is nothing to export into: take the path it printed and substitute it literally wherever these instructions write $CLAUDE_PLUGIN_ROOT. Never guess it.
Printed NOT FOUND? Stop, and tell the user to reinstall the plugin with /plugin marketplace add storehand/storehand.
Read $CLAUDE_PLUGIN_ROOT/shared/safety.md before you start.
Step 1 — Check the Shopify CLI
shopify version
- Not found: tell the user to install it (
npm install -g @shopify/cli) or
see https://shopify.dev/docs/api/shopify-cli, then stop. Do not continue.
- Below 4.5: ask them to run
shopify upgrade.shopify store executeis
required and does not exist in older versions.
Step 2 — Check for an existing profile
Look for .storehand/store.yaml in the working directory.
- Exists: show the user what is in it and ask whether they want to update it
or set up a different store. Never silently overwrite a profile.
- Does not exist: continue.
Step 3 — Ask for the essentials
Ask these one at a time. Do not fill in guesses.
- The
*.myshopify.comdomain (not their custom domain — the admin URL shows
it). If they give a custom domain, ask again for the myshopify one.
- Timezone and currency. Offer a sensible guess based on the domain or their
language and let them confirm.
- The language the store sells in, as an ISO code (
nl,en,de).
Offer a guess from the domain and let them confirm — but get an answer, do not leave it blank. Everything StoreHand writes goes in this language, and text it finds in another one becomes a finding.
Ask it even when the answer looks obvious. A store importing from a foreign supplier is full of the supplier's language, so a skill that copied "whatever is already there" would keep writing it — measured on a live Dutch store carrying 429 English alt texts.
- The low-stock threshold: "below how many items do you want a warning?"
- The minimum margin percentage StoreHand must never price below. **Optional
and not read by any shipped skill yet** — if they don't know, leave it null rather than inventing a number.
Step 4 — Write the profile
Create .storehand/ and copy the templates from this plugin, filling in the answers:
store.yamlfrom$CLAUDE_PLUGIN_ROOT/templates/storehand/store.yaml—
replace every placeholder
store.mdfrom$CLAUDE_PLUGIN_ROOT/templates/storehand/store.md— leave the
comment prompts in place so the user can fill them in later
Then tell the user plainly: store.md is what makes the writing skills sound like their store. It is worth ten minutes of their time, but it is not needed for the daily briefing, so they can do it later.
Step 5 — Connect the store
Show the command, explain it, then run it:
shopify store auth --store --scopes read_orders,read_products,read_inventory,read_discounts,read_online_store_navigation
Explain before running:
- This uses Shopify's own CLI app. The user does not create an app, does not
need a Partner account, and there is no app review.
- A browser opens and shows exactly which permissions are being requested. These
five scopes are read-only — StoreHand cannot change anything with them.
- The token is stored locally by the Shopify CLI in
~/.config/, and it refreshes
itself.
If no browser is available (a server, a remote SSH session), the CLI still needs one — it has no device-code flow. Verified against CLI 4.5.2 on 2026-08-03: SHOPIFY_CLI_DEVICE_AUTH is listed among the CLI's environment variable names but is read nowhere in its code, so setting it changes nothing and shopify store auth still fails with spawn xdg-open ENOENT.
What works is borrowing the browser on another machine and tunnelling the callback back. Three steps, in this order:
- Give the headless box something to "open" with, so the URL is printed
instead of the command failing:
``bash printf '#!/bin/sh\nprintf "%%s\\n" "$1"\n' > /usr/local/bin/xdg-open chmod +x /usr/local/bin/xdg-open ``
- From the machine that does have a browser, open a tunnel and leave it
running. Read the port out of the printed URL rather than assuming it — it is the redirect_uri parameter, and it is not the 3456 that appears as a constant in the CLI's own source:
``bash ssh -L :127.0.0.1: @ ``
- Run
shopify store authon the headless box. It prints the authorization
URL. Open that URL in the browser on the other machine and approve. Shopify redirects to 127.0.0.1:, the tunnel carries it back, and the CLI stores the token.
If the tunnel was not up in time, the browser shows a connection error but the authorization code is still in its address bar and the CLI is still listening. Deliver it by hand on the headless box — the code is single use, so do this once:
curl "http://127.0.0.1:/auth/callback?code=&shop=&state="
Note on order history: the read_orders scope covers roughly the last 60 days, which is more than a daily briefing needs. Longer history needs extra permission from Shopify and is not part of StoreHand.
If they want the listing writer
product-listing-writer is the only skill that changes anything, and it needs one scope more. Offer this, do not run it by default — a store connected with the five read-only scopes is the safer resting state, and every other skill works without it.
shopify store auth --store --scopes read_orders,read_products,read_inventory,read_discounts,read_online_store_navigation,write_products,write_files
Say what it does and does not buy them:
write_productslets StoreHand change product titles, descriptions, SEO
fields and image alt text. Nothing else — not prices, not inventory, not orders, not customers.
- Even with the scope granted, nothing is written without them approving that
specific change set first. The listing writer proposes into a file they read and edit, and applies only what is still in it.
- Without the scope, the listing writer still works up to the point of writing:
it produces the proposal and then stops, saying which scope is missing.
They can add it later. Re-running the auth command with the longer scope list replaces the old grant, so this is not a decision they are stuck with.
Step 6 — Smoke test
shopify store execute --store --json --query '{ shop { name currencyCode timezoneAbbreviation } }'
- Works: show the store name and confirm it is the right store. If the
currency or timezone differs from what they told you, point that out and offer to correct store.yaml.
- Fails: show the exact error. Common causes: wrong domain, the account
lacks permission to grant app access, or the auth step never finished. Do not claim setup succeeded.
Step 7 — Hand off
Confirm what exists now (.storehand/store.yaml, .storehand/store.md, a connected store) and suggest the obvious next move: run the daily-store-briefing skill.
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: storehand
- Source: storehand/storehand
- License: Apache-2.0
- Homepage: https://storehand.github.io
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet, be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.