shopify-skills
CLAUDE.md1 starsChanged 20 days agoMITView on GitHub
Skills live at skills/<bucket>/<skill-name>/SKILL.md. Buckets: support/ (talking to customers), store/ (the storefront and catalog) and operations/ (profit, deliveries, returns, stock). When you add, rename or remove a skill, update all four in the same change: Then run claude plugin validate . (skip --strict: it fails on this file sitting at the plugin root, which is meant for maintainers, not plugin users). The frontmatter name must match the folder name. Keep description under 200 characters, because older Claude.ai builds reject longer…
In this repository12 other files it gives its agents.
- cod-rto-analysisSkill
- dead-stock-clearanceSkill
- demand-plannerSkill
- returns-reason-analysisSkill
- true-profitSkill
- weekly-kpi-digestSkill
- catalog-photoshootSkill
- conversion-optimiserSkill
- product-descriptionsSkill
- shopify-catalog-uploadSkill
- size-chart-builderSkill
- angry-customer-replySkill
The file
We split this CLAUDE.md into the rules it gives an agent
4 rules. About 0 are common, 3 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 5
When you add, rename or remove a skill, update all four in the same change:
- Distinctive: line 10
Remove it from "Coming Soon" in README.md if it was listed there.
- Very specific: line 30
Read both spellings of renamed Shopify export headers: Cancelled at/Canceled at, Handle/URL handle, Variant SKU/SKU, Option1 Value/Option1 …
- Distinctive: line 36
Never put real merchant data in tests or fixtures.
≈ “Common” is an estimate: at least 1 in 5 of the 28,029 CLAUDE.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Maintaining this repo
Skills live at skills/<bucket>/<skill-name>/SKILL.md. Buckets: support/ (talking to customers), store/ (the storefront and catalog) and operations/ (profit, deliveries, returns, stock).
When you add, rename or remove a skill, update all four in the same change:
- The top-level
README.md: the download link in the Claude install block, and the entry under Reference, with the skill name linked to itsSKILL.md. - The bucket's
README.md. - The
skillsarray in.claude-plugin/plugin.json. - Remove it from "Coming Soon" in
README.mdif it was listed there.
Then run claude plugin validate . (skip --strict: it fails on this file sitting at the plugin root, which is meant for maintainers, not plugin users).
The frontmatter name must match the folder name. Keep description under 200 characters, because older Claude.ai builds reject longer ones.
The readers are shop owners, not developers. Use plain words in skills and in what they tell the model to write. No em dashes in any prose: rewrite the sentence with a comma, colon, full stop or parentheses instead.
To release, bump version in .claude-plugin/plugin.json, commit, then push a tag vX.Y.Z. The Release workflow attaches one zip per skill, which is what the README's download links point at.
The skills pack (store and operations skills with scripts)
The ten skills listed in pdf/content.json are the free skills pack: each has a first-run setup, a Safety section, Python scripts and tests. Read CONVENTIONS.md (pack rules) and SPECS.md (locked scope) before changing one.
python3 tools/lint.pychecks every skill (spec, description length, dashes, private leftovers, file references; pack rules for the pack skills).python3 -m unittest discover testsruns the tests. Scripts are Python 3.9+ standard library only.shared/tabular.pyis copied verbatim into each pack skill'sscripts/; edit the shared copy and re-copy.python3 pdf/build.pyrebuilds the lead magnet PDFs intopdf/out/(git ignored). The install line reads each skill's bucket folder from the repo, so rebuild after moving a skill.
Invariants (rules learned from bugs)
- Read both spellings of renamed Shopify export headers:
Cancelled at/Canceled at,Handle/URL handle,Variant SKU/SKU,Option1 Value/Option1 value/Option 1 Value. Shopify renamed them in 2026 exports and the skills silently misread orders. - A missing unit cost is unknown, never 0. Showing "EGP 0" or a profit number without costs misleads the merchant.
- On COD stores a failed delivery is usually "fulfilled + cancelled + voided", not a tag. Every skill must use the same failed/delivered rules as
cod-rto-analysis. - Status words must be matched as negation-aware phrases: "لم يتم التسليم" contains "تم التسليم", "Out for delivery" contains "deliver".
Publication.nameis deprecated (API 2026-10) and connectors refuse deprecated fields: usecatalog { title }.- Magnific Seedream 5 Pro has no 4:5 ratio; house-model candidate prompts must force the full head in frame.
- Never put real merchant data in tests or fixtures.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

