Product SEO Fixer

Store: a beauty brand on Shopify (name withheld). Real public product data, converted to Shopify's export format for the test; we're not affiliated. Product and shade names are replaced with placeholders.

Before: 25 products, 159 images. 157 images had no alt text. One 3-piece lip set ("Set A") was showing another set's SEO title and meta description, word for word, copy-pasted from "Set B".

After (rows from seo-fixes.csv, test 06):

Set A got its own SEO title ([Set A] | [Line] 3-Piece Lip Stain & Gloss Set) and a meta naming its own three shades. All 159 alt texts and 5 SEO fixes came in one 6-column CSV. Alt text is built only from real data (product name, linked variant, filename hints), so where the image content is unknown it says "view N" rather than making up a scene.

Collection example (test 03, a home-textiles brand): the bath collection's meta description was the store's bedding copy, and the page was noindex. The skill flagged the noindex as probably deliberate and pointed the fixes at the indexable twin page.

Finds the on-page SEO problems you can fix in Shopify admin and hands you the exact text to paste, or a CSV to import. Every finding quotes real data from your store; nothing is generic advice.

Tested on 5 real public Shopify stores across all three modes. Final runs scored 9 to 10 out of 10.

Tested on an apparel brand, a beauty brand, a home-textiles brand and a footwear brand (CSV), all real public Shopify stores. 7 final tests covering all three modes, scored 9, 10, 9, 10, 10, 10, 10 out of 10.

Update 2026-10-04: collection audits now copy their counts and product lists (empty, thin and duplicate descriptions, missing alt text, word and character counts) from the script output, and collection copy always speaks as the store (we / our). Re-tested on a food brand's collection (two runs) and re-run on one earlier collection test and one product test: 10, 9, 9.5 and 9 out of 10, with every count and handle matching the fetched data.

Update 2026-10-04 (second check): run on a product with almost no description. The first run added benefit wording the product page did not state; the skill now writes only what the source says and a re-run came back clean. One run on one product, not scored.

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 sees server HTML only. Confirm structured data with Google's Rich Results Test.
• Image alt text is drafted without seeing the images (from names, variants and filenames). Spot-check a few.
• Shopify's 2024+ export headers are supported by name mapping but haven't been tested on a real new-format export yet.

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 helper scripts need Python 3 (standard library only). Reading a live store needs web access.

1. Unzip the download. It holds the product-seo-fixer folder, ready-made zips for claude.ai in a claude-ai folder, a README and the licence.
2. Put the product-seo-fixer 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: "Audit the SEO of https://yourstore.com/products/xyz and give me the fixes". The skill loads by itself.