Support Macro Writer

Before: a footwear brand's refund policy says, among 900+ words, "Returns must be initiated within 30 days of purchase" and "A flat-rate fee of $10 will be deducted from refunds". Its FAQ separately says a gift-card order "has to be returned, a refund will not be possible", while the policy page says the refund "will also be issued via gift card".

After (excerpt from the skill's output, unedited):

Gift-card orders: Policy says the refund goes to a gift card. FAQ says 'a refund will not be possible', exchange only. Affects M12, M15. Pick one.

M06 · Cancel after fulfilment
Before sending: shipped and tracking status; final sale items; purchase date vs the 30 days; payment method.
Hi {{customer.first_name}}, {{order.name}} has already shipped, so I can't cancel it. Once it arrives, you can return it within 30 days of purchase if it's unworn with original tags and packaging, using our returns portal: [DECIDE: D1 returns portal URL]. A flat-rate fee of $10 is deducted from refunds. Final sale items can't be returned. [DECIDE: D6 can the customer refuse the parcel?]

The same run also flagged the missing rules (lost parcels, duties, cancellation and address-change windows, among others) and turned each one into a one-time decision with a suggested default.

Give it your store URL. It reads your real refund and shipping policies and writes the support macros your team actually needs. It starts with the exception cases that eat the day:

1. Parcel stuck in transit
2. Marked delivered, not received
3. Address change before and after fulfilment
4. Cancel before and after fulfilment
5. Partial or split shipment
6. Customs or duties surprise
7. Damaged item or wrong item (with a photo request)
8. Processing delay
9. Returns inside and outside the window, exchanges, refund status
10. Discount code not working
11. The plain "where is my order" reply, last

Every macro has:

• an agent checklist (what to check in Shopify admin or with the carrier before sending);
• the policy line it relies on, quoted;
• a customer reply under 120 words with one clear next step.

It also gives you:

• A policy facts sheet, with every rule quoted from your own pages.
• Contradictions and gaps, e.g. a refund page that says 60 days while the FAQ says 30, or no rule at all for lost parcels. Where your policy is silent, it writes [DECIDE: …] instead of inventing a rule, and lists every decision with a suggested default.
• An import pack: variables in Gorgias or Zendesk syntax ({{customer.first_name}}-style), Shopify Inbox saved-reply versions, a CSV, and Zendesk API JSON. The bundled script macros_to_import.py turns the pack into a correctly quoted CSV.

Tested on 4 real public Shopify stores with Gorgias, Zendesk, Shopify Inbox and neutral output, scored against a written rubric. Every policy fact in the macros was checked against the stores' own policy text.

The tests ran on real public data, never made-up cases, and were scored against a written rubric. A skill in this range ships only when every test scores 8/10 or more.

How we test, what we do not do, and which assistants we ran it on.

• It only knows what your pages say. Help centres that load only with JavaScript (some Gladly or Kustomer setups) can't be read. Paste those articles and it fills in the matching decisions.
• Variable names for Gorgias' Shopify fields follow Shopify's order fields. Insert one with your helpdesk's variable picker to confirm before a bulk import.
• It writes no legal terms. Where you ship to the EU, UK or Australia, it reminds you to check local consumer-law minimums.

Works with any assistant that reads SKILL.md skills (Claude, Codex, Gemini CLI, Cursor and others), plus a paste-in prompt version for ChatGPT or any chatbot. We run our release tests on Claude Sonnet and Opus. Use your assistant's strongest model: in our tests, a small, fast model added facts that were not in the source data. The fetch and import scripts need Python 3 (standard library only) and web access. Without them, paste your policy text and the skill works from that.

1. Unzip the download. It holds the support-macro-writer folder, ready-made zips for claude.ai in a claude-ai folder, a README and the licence.
2. Put the support-macro-writer folder where your assistant looks for skills. The generic path is .agents/skills/ (or npx skills add from a folder). Tool-specific paths:
   • Claude Code: ~/.claude/skills/ (or .claude/skills/ inside one project)
   • OpenAI Codex, Gemini CLI, Cursor and others: .agents/skills/ or the skills folder their docs name
   • claude.ai: Customize, Skills, then the plus button, Create skill, Upload a skill, and upload the zip from the claude-ai folder without unzipping it (code execution must be on in Settings, Capabilities)
   • ChatGPT or any chatbot: open the skill's prompt.md, paste it into the chat, then paste your own text or link
   • Steps for each tool, with links to its official docs: how to install Agent Skills
3. Ask for the job in plain words, for example: "Write support macros for our Gorgias based on our policies: mystore.com". The skill loads by itself.