Codex reads skills from .agents/skills in your repository and from $HOME/.agents/skills for every repository, per OpenAI's docs. Copy a skill's folder there, or run npx skills add proskillpacks/skills -a codex. Restart Codex if a new skill does not appear.
Step-by-step install page for Codex CLI, with the official docs it comes from.
Tested on
Our release tests ran on Claude Sonnet, and on Claude Opus for some skills; each skill page says what it was tested on. We do not claim a separate test in Codex CLI. The skills are plain files in the open SKILL.md format, so any assistant that reads that format should be able to load them, but we have not verified each tool. How we test.
Free skills
Free, MIT licence, on GitHub. Install with npx skills add proskillpacks/skills.
AI Search Readiness Check
Can ChatGPT, Perplexity and Google read a website? A score out of 100 from 8 checks, the evidence and the fixes in order.
Tested on 8 real public websites, as a skill and (four of them) as a pasted prompt, two also on a second model, then re-scored after two independent reviews.
Accessibility Quick Audit
Give it a page's HTML or components. Get accessibility issues ranked by impact, with the WCAG criterion, the code and a fix. Free, MIT-licensed.
1 real-data run on a public demo page built with known accessibility faults (Claude Sonnet).
Agent-Ready Quick Check
Six checks on how AI shopping agents read your store, compared with 99 stores we audited.
Tested on 4 real public Shopify stores.
Alt Text Writer
Alt text for every product image, written while looking at the photo, as an import CSV.
Tested on 6 real cases from public Shopify stores, including three product export CSVs.
Angry Customer Reply Check
Give it the complaint, your draft reply and your policy. Get a check of promises, facts and tone, and a repaired draft. Free, MIT-licensed.
1 run on a complaint, draft and policy we wrote (Claude Sonnet).
CSV Profile and Sanity Check
Give it a CSV. Get a read-only profile of every column, flags and questions for the data's owner. Free, MIT-licensed.
1 run on a real public CSV (Claude Sonnet) with the script.
Changelog from Commits
Give it a commit range or a pasted git log. Get a user-facing changelog with the commit behind every line, and the vague commits flagged.
1 real-data run on the public Express repository (Claude Sonnet, 15 commits).
Chargeback Evidence Checklist
Got a chargeback? The evidence to gather for your dispute reason, what you already have, and where to find each piece in Shopify.
Tested on 6 real disputes from the Shopify Community forum.
Cold Email Checker
Give it a cold email draft and the prospect's page text. Get a verdict, a claims table and a tightened version. Free, MIT-licensed.
1 run on a draft we wrote with common flaws and text from a real public homepage (Claude Sonnet).
Fact-Claim Checker
Paste a draft before you publish. Get the claims that need a source, ranked by risk, what would settle each, and safer wording for the ones you cannot back up.
1 real-data run on a public launch post (Claude Sonnet).
Invoice Checker
Give it an invoice. Get the totals recomputed, every difference, missing fields and consistency flags. No tax advice. Free, MIT-licensed.
1 run on a sample invoice we wrote with planted errors (Claude Sonnet).
Job Post Decoder
Paste a job post. Get what the job is, firm and wish requirements, what is missing and questions for the recruiter. Free, MIT-licensed.
1 run on a real public job post (Claude Sonnet).
Late Payment Chaser (UK)
One unpaid invoice in. The statutory interest and fixed recovery sum worked out with a script, and three reminder emails out. UK business debts only.
3 real tests built on the facts of published UK court judgments.
PR Description and Review Prep
Run it on your branch. Get a PR title, summary, risk notes and test plan from the diff, plus the questions a reviewer will ask. Free, MIT-licensed.
1 real-data run on four public commits (Claude Sonnet).
Policy Checker
Policy gaps and contradictions across refund, shipping, privacy and terms, with fixes to paste.
Tested on 4 real public Shopify stores.
Product Title Cleaner
One consistent title pattern across the catalogue, as an import CSV of changed rows only.
Tested on 3 real public Shopify catalogues, up to 490 products.
README First-Run Check
Point it at a repo. It follows the README like a new user and checks every step against the repo: OK, BROKEN, MISSING or UNVERIFIED, with evidence and the README edits that fix it.
1 real-data run on a public MDN tutorial repository (Claude Sonnet).
Review Reply Drafter
A reply for every review, matched to its tone, with the risky ones flagged for you.
Tested on 3 real public review sets (Judge.me and Trustpilot).
Search Intent Page Brief
Give it a keyword and the pages that rank. Get the search intent, a coverage table, the gaps and an outline. Free, MIT-licensed.
1 real-data run on three public how-to pages (Claude Sonnet).
Skill Portability Check
A free checker for SKILL.md folders: strict-YAML trap, portable keys, vendor tool names, missing fallbacks, broken links. One Python file, exit code for CI.
Run over our own 57 skills and its four fixtures (5 checks pass).
Test Gap Finder
Give it a diff and the tests. Get each behaviour the change adds marked Covered, Not covered or Unknown, and ranked test cases for the gaps.
1 real-data run on a public multi-hunk change in the Express repository (Claude Sonnet).
Free prompts
Copy, paste into any chatbot, add your own material. No install, no sign-up.
- AI search readiness check prompt for any chatbot: Can ChatGPT, Perplexity and Google read a website? A score out of 100 from 8 checks, the evidence and the fixes in order.
- Accessibility audit prompt for ChatGPT, Claude or Gemini: Give it a page's HTML or components. Get accessibility issues ranked by impact, with the WCAG criterion, the code and a fix. Free, MIT-licensed.
- Can AI agents read my Shopify store? Free prompt: Six checks on how AI shopping agents read your store, compared with 99 stores we audited.
- Product image alt text prompt for ChatGPT or Claude: Alt text for every product image, written while looking at the photo, as an import CSV.
- Angry customer reply checker prompt for any chatbot: Give it the complaint, your draft reply and your policy. Get a check of promises, facts and tone, and a repaired draft. Free, MIT-licensed.
- CSV sanity check prompt for ChatGPT or Claude: Give it a CSV. Get a read-only profile of every column, flags and questions for the data's owner. Free, MIT-licensed.
- Changelog from commits prompt for ChatGPT or Claude: Give it a commit range or a pasted git log. Get a user-facing changelog with the commit behind every line, and the vague commits flagged.
- Chargeback evidence checklist prompt for ChatGPT or Claude: Got a chargeback? The evidence to gather for your dispute reason, what you already have, and where to find each piece in Shopify.
- Cold email checker prompt for ChatGPT, Claude or Gemini: Give it a cold email draft and the prospect's page text. Get a verdict, a claims table and a tightened version. Free, MIT-licensed.
- Fact-check a draft: claim checker prompt for any chatbot: Paste a draft before you publish. Get the claims that need a source, ranked by risk, what would settle each, and safer wording for the ones you cannot back up.
- Invoice checker prompt for ChatGPT, Claude or Gemini: Give it an invoice. Get the totals recomputed, every difference, missing fields and consistency flags. No tax advice. Free, MIT-licensed.
- Job post decoder prompt for ChatGPT, Claude or Gemini: Paste a job post. Get what the job is, firm and wish requirements, what is missing and questions for the recruiter. Free, MIT-licensed.
- Late payment interest and chaser email prompt (UK): One unpaid invoice in. The statutory interest and fixed recovery sum worked out with a script, and three reminder emails out. UK business debts only.
- Pull request description prompt for ChatGPT or Claude: Run it on your branch. Get a PR title, summary, risk notes and test plan from the diff, plus the questions a reviewer will ask. Free, MIT-licensed.
- Refund policy checker prompt for ChatGPT, Claude or Gemini: Policy gaps and contradictions across refund, shipping, privacy and terms, with fixes to paste.
- Shopify product title cleaner prompt for any chatbot: One consistent title pattern across the catalogue, as an import CSV of changed rows only.
- README first-run check prompt for any chatbot: Point it at a repo. It follows the README like a new user and checks every step against the repo: OK, BROKEN, MISSING or UNVERIFIED, with evidence and the README edits that fix it.
- Review reply drafter prompt for ChatGPT, Claude or Gemini: A reply for every review, matched to its tone, with the risky ones flagged for you.
- Search intent content brief prompt for any chatbot: Give it a keyword and the pages that rank. Get the search intent, a coverage table, the gaps and an outline. Free, MIT-licensed.
- Check a SKILL.md for portability problems: free prompt: A free checker for SKILL.md folders: strict-YAML trap, portable keys, vendor tool names, missing fallbacks, broken links. One Python file, exit code for CI.
- Find missing tests: test gap finder prompt for any chatbot: Give it a diff and the tests. Get each behaviour the change adds marked Covered, Not covered or Unknown, and ranked test cases for the gaps.
Paid skills and packs
Paid skills and packs, one-time payment on Gumroad.
Shopify store owners
Store Ops Pack
Six skills for the Shopify jobs that eat the week, in one download.
Every skill in the pack was tested on real public Shopify stores or real public merchant cases and scored against a written rubric.
Agent-Ready Store Audit
A 0 to 100 score for how AI shopping agents read your store, with the top 7 fixes.
Tested on 4 real public Shopify stores.
BFCM Campaign Kit
Offer maths on your real prices, a dated calendar, 5 emails, 3 SMS and an ops checklist.
Tested on 4 real public Shopify stores (US and UK, low to high price points).
Chargeback Response Kit for Shopify Payments, with notes for PayPal, Stripe and Klarna
Six fill-in response letters by reason, evidence checklists, worked examples and a Start-here PDF, plus an AI prompt and skill. One price, $5.
Tested on 6 real dispute scenarios from the Shopify Community forum.
Product Description Rewriter
Clean Shopify HTML descriptions with an SEO title and meta description, for one product or a collection.
Tested on 4 real public Shopify stores and one product export CSV.
Product SEO Fixer
A prioritised fix list with the exact text to paste, for a product, a collection or a product export CSV.
Tested on 5 real public Shopify stores across all three modes.
Support Macro Writer
Ready-to-import replies built from your real shipping and returns policies.
Tested on 4 real public Shopify stores with Gorgias, Zendesk, Shopify Inbox and neutral output, scored against a written rubric.
Short-term rental hosts
Host Ops Pack
Six skills for the listing, review and house-rules jobs of hosts who run their own listings.
Each of the six skills has 4 tests on real public data, scored by hand against a written rubric.
Listing Rewriter
Three titles under 50 characters and a description within 500, built only from your own facts.
Tested on 4 real public listings.
Follow for new free skills
New free skills and guides appear in all of these as we publish them. Pick whichever you already use.
- RSS feed for any feed reader
- Watch the free skills repo on GitHub
- Follow on Gumroad