agentleFS
OutThisLifeOutThisLife/brooklyn-skills/skills/shop-board/SKILL.md

shop-board

Skill215 starsChanged 4 months agoMITView on GitHub

Turn any online shop into a calm, visual board in the browser: the whole catalog as a photo grid, filtered to your sizes, grouped by intent instead of product type, click to pick, then add the picks to the store's cart. Also a click-to-remove review view for a cart that's already full. Takes loose args like "/shop-board dollskill xs, xxs, cutie stuff only". Use for "make a board for <shop>", "go through <store> for me", "find me stuff on <store>", "review my cart", or any browse-a-whole-store / trim-a-cart ask.

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 1 is distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Anything else: run scripts/crawl-listing.js first.

    line 51
  2. Distinctive:

    WooCommerce removes are batched through the cart form: Every write re-totals the whole cart: one removal at a time takes seconds per item …

    line 81
  3. Very specific:

    wc-ajax=remove_from_cart returns 200 even when it fails: Only a response with fragments actually removed the item.

    line 85
  4. Very specific:

    When someone settles on a cart, save it as [name, size] pairs so it can be rebuilt, and suggest they log in.

    line 92

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill 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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Shop board

A shop's own pages are built to sell one product at a time. The board puts the whole catalog on one screen: photos, in stock in the right size, sorted into groups that match what the person is shopping for. Click to pick, and the picks go to the cart.

It runs in the person's own browser session through a browser-automation tool that can execute JavaScript in a page (Claude in Chrome, Playwright and similar). Scripts are pasted whole into the page. Nothing is installed, and nothing is bought: the deliverable is a cart.

Read the request

references/vibes.md maps loose words to settings: the store, sizes (optional), vibe groups, "… only", promo codes. No sizes means every size. No vibe words means the general groups (Everyday, Going out, Lounge & sleep, Intimates, Statement, Active & swim, Outerwear, Shoes, Bags & accessories).

Flow

  1. Open the store in a new tab and work out the platform. If window.Shopify exists or /products.json returns JSON, it's Shopify. If form.cart, woocommerce classes or /?wc-ajax= exist, it's WooCommerce. Anything else is a custom site.
  2. Check the size labels before filtering. Count the size option's values on the first page of products. Stores spell sizes out ("X-SMALL", "XX-SMALL", "X-SMALL/SMALL"). board.js handles the common aliases, but a store with its own scheme needs sizes written in its words.
  3. Look for house brands. A store's own lines show up as collections, and their keywords won't catch them. Pass them as brands. They're crawled whole and get a switch at the top.
  4. Open the board. Set window.__board, then paste scripts/board.js.
    • Shopify: it crawls products.json. That feed stops at page 100 (25k products), so /collections/all is crawled too and merged by id. Progress is saved per store and per size set, so a closed tab resumes and a finished crawl reopens instantly. Add to cart goes through /cart/add.js in the smallest wanted size. Shoes are flagged for the person to choose a size.
    • Anything else: run scripts/crawl-listing.js first. Its defaults are WooCommerce. For a custom site, inspect one product card and set the selectors. Then set window.__board.items = window.__items. The board opens as a shortlist: picks are saved, and the person adds to the bag from each product's View link. Say so when the listing shows no stock or sizes.
    • WooCommerce add-to-cart: scripts/woo-add.js with the picked URLs.
  5. Review a full cart. scripts/cart-review.js on any store page. Click to mark for removal, Undo reverses a click, Recent shows what changed, and Apply removes the marked items.
  6. Report: what's on the board (count, groups), what was added or removed, what got skipped and why. Say that nothing was checked out and which tab it's in.

Long jobs (crawls, multi-item adds) run in the background and write progress to a window.__… object. Poll it with short waits. Page evaluation times out at around 45s, so never await a whole crawl in one call.

Hard-won rules

  • Keep it fast on big stores. Request CDN thumbnails (Shopify resizes with ?width=400; originals run 1–4k px and stutter scrolling), let content-visibility skip offscreen cards, avoid live backdrop-filter over the grid, and update one card on a pick instead of re-rendering hundreds.
  • Nothing vanishes under the cursor. A click toggles one thing. A card stays where it is until the tab or group changes. Bulk actions (Keep all / Remove all) come with Undo.
  • Marks are the source of truth. Clear a mark only after the cart confirms the item is gone, so an interrupted Apply loses nothing.
  • WooCommerce adds are sequential, about 800ms apart. Parallel or back-to-back POSTs race the session and silently drop about every other item. Always diff the cart afterward.
  • WooCommerce removes are batched through the cart form. Every write re-totals the whole cart: one removal at a time takes seconds per item on a big cart, and one giant request 502s at the gateway. Zero out about 6 lines per request and halve the chunk on any non-200.
  • wc-ajax=remove_from_cart returns 200 even when it fails. Only a response with fragments actually removed the item.
  • Read the cart with a cache-busting query (/cart/?r=<now>). Chrome can serve a stale cart page, and a retry based on it doubles the item.
  • Variant labels may only exist as codes (lb-sn802-s301 for "Black"). Match variants by the code that was picked, not by the label.
  • Guest carts expire (WooCommerce keeps a guest session for days, not weeks). When someone settles on a cart, save it as [name, size] pairs so it can be rebuilt, and suggest they log in. They log in themselves; never enter credentials.
  • The overlay goes after <body> (documentElement.appendChild). That keeps same-z-index cookie banners from sitting on top of it. Decline non-essential cookies if a banner blocks the view.
  • Close tabs you opened; never repurpose the person's own tabs. Navigating a tab wipes window state, which is why crawls persist to localStorage.

Look

Both views use shadcn's zinc tokens at a compact density and follow the system light/dark setting. There's no theme switch and no sticky header. The views live in a shadow root, so the store's CSS can't leak in.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.