ship-options-il

name	ship-options-il
description	Enumerate and evaluate every shipping option a listing offers to Israel — carrier lane, cost, lead time, tracking, tax-at-checkout — and recommend one. The logistics half of a buy decision, which the search skills deliberately leave out.

Shipping Options to Israel

For one listing (or each of a shortlist), open the delivery panel, read every shipping option offered to Israel, and evaluate them against each other rather than accepting the default the site pre-selects.

Why this is its own skill

The search skills return a price. The price is the smaller half of the decision for an Israeli buyer:

• The collapsed shipping row is not the menu. AliExpress pre-selects one option — usually the cheapest, frequently a 30–60 day lane. A ₪4 upgrade often halves the wait. You cannot see that without opening the panel.
• "Free shipping" says nothing about speed. Free is routinely the slowest carrier. See free-shipping-only, which filters on the flag but explicitly does not evaluate the lane.
• Ship-from country dominates lead time. An IL-warehouse listing arrives in days; the identical item from CN takes weeks. The search filter exposes this but the cards do not always show it.
• Choice is consolidated logistics, which changes both the delivery estimate and how customs treats the parcel — Choice items in one order likely clear as one consignment. That matters to cart-vat-nudge's split-order advice.
• Shipping cost feeds the tax calculation. Once over $75 the VAT is charged on CIF, so the shipping figure is not a rounding error — see $CLAUDE_PLUGIN_ROOT/reference/israel-tax.md.

When to use

• User asks "how long will this take to get here", "what are my shipping options", "is it worth paying for faster", "does this actually ship to Israel"
• Between search-aliexpress and a buy decision, on the two or three finalists
• Before find-under-75, when the landed figure depends on which lane is chosen
• When a listing shows free shipping and the user wants to know the catch

Inputs

• url (required) — listing URL, or an item id. Accepts a list for shortlist mode.
• sku (optional) — variant to price. Shipping can differ per variant (weight bands); if omitted, use the default and say so.
• qty (optional, default 1) — shipping frequently does not scale linearly, and some lanes only unlock above a subtotal.
• priority (optional, default balanced) — cheapest | fastest | balanced | tracked. Drives the recommendation, not the extraction.

Browser route

Chrome first. See $CLAUDE_PLUGIN_ROOT/reference/browser.md for the route order, tool mapping and the ILS/Hebrew/ship-to-IL locale handshake — do the handshake before reading any figure, because the whole panel is keyed on ship-to country.

Confirm ship-to is Israel from the rendered page, not from an assumption:

document.body.innerText.match(/ישראל|Israel/)?.[0] ?? 'SHIP-TO NOT CONFIRMED'

If ship-to is not Israel, the options listed belong to another country. Stop, fix the region, re-read.

Extraction

Selectors for this panel are not yet validated — see the (U) section of $CLAUDE_PLUGIN_ROOT/reference/selectors.md. Work text-first, and promote what you find into that file with a date.

1. Open the panel

The collapsed row shows one option. Find and click the shipping row to expand the full list:

// Text anchor — most durable. Returns the clickable shipping row.
const row = [...document.querySelectorAll('div,button,a')]
  .filter(el => /^(משלוח|Shipping|אספקה|Delivery)/i.test(el.textContent.trim()))
  .sort((a, b) => a.textContent.length - b.textContent.length)[0];

Click it with computer/browser_click on the element's box rather than a synthetic .click() where possible — the panel is sometimes gated on a real pointer event. A modal ([class*="comet-v2-modal"]) or an inline expansion appears.

2. Read every option

Each option row carries some subset of: carrier/lane name, cost (or "חינם" / "Free"), estimated delivery window, and a tracking indicator. Extract as text and parse:

[...document.querySelectorAll('[class*="comet-v2-modal"] [class*="item"], [class*="dynamic-shipping"] [class*="line"]')]
  .map(el => el.innerText.trim())
  .filter(Boolean);

Then parse each blob for:

Field	Cue
carrier	lane name — AliExpress Standard, Choice, Cainiao, a courier brand
cost_ils	₪ amount, or free (חינם / Free)
lead_min / lead_max	day range — 10-30, 10–30 ימים, or a delivery date
tracked	presence of tracking wording; absence is not proof of untracked
ship_from	origin country if the panel states one

Do not invent a field the panel did not show. Emit null and mark it unknown. A guessed lead time on a logistics decision is worse than a blank.

3. Capture procedure (run once, then update the reference)

The first time this skill runs against the live site, record what actually worked: the selector that found the row, the selector that enumerated options, and one raw innerText sample of an option row. Write them into the (U) section of reference/selectors.md, change the status to V, and date it. That is the whole point of the section existing.

Evaluation

Rank the options; do not just relay them.

1. Normalize — every option to {cost_ils, cost_usd, lead_min, lead_max}. Use the shared FX rule (reference/israel-tax.md): live frankfurter rate, print the rate, stop if unavailable.
2. Compute the upgrade cost of time — for each non-cheapest option, ₪ per day saved = (cost − cheapest_cost) / (cheapest_lead_mid − this_lead_mid). This is the number that actually answers "is faster worth it".
3. Flag the structural facts, which matter more than small cost deltas:
   • shipping from IL → days not weeks; usually worth a premium
   • Choice → consolidated, generally faster and more reliably tracked than a seller-arranged lane at similar cost
   • a lane with no tracking on a parcel worth real money is a risk, not a saving
   • an option whose window exceeds the buyer-protection period is a trap — the protection can lapse before the parcel lands
4. Recommend one, per priority, in a sentence with the reason.
5. Hand the chosen shipping cost onward to fetch-listing / find-under-75 / cart-vat-nudge — the landed figure is only meaningful for a specific lane.

Output format

Listing: <title>   item <id>
Ship-to: Israel (confirmed on page)   Locale: ILS verified
Variant: <sku or "default (not selected)">   Qty: <n>
Route: claude-in-chrome        FX: 1 USD = ₪<rate> (<source>@<date>)

Shipping options to IL (<N> found):
  carrier                    cost      lead        tracked
  AliExpress Standard        ₪7.49     10–30 d     yes
  Choice consolidated        free      12–18 d     yes
  <lane>                     ₪24.10    5–9 d       yes
                                                   ← unknown fields shown as "?"

Cost of time:
  Choice → <lane>:  ₪24.10 for ~8 days saved  =  ₪3.01/day

Structural notes:
  • ships from CN — no IL-warehouse option on this listing
  • Choice parcel: consolidates with other Choice items in the same order
  • <lane> window (5–9 d) is inside buyer protection; Standard's 30 d upper is not

Recommendation (priority=balanced):
  Choice consolidated — free, tracked, and only ~6 days behind the paid lane.

Landed at this lane: item ₪<x> + ship ₪<y> = ₪<z>  ($<usd>)   VAT band: <…>
Next: `find-under-75` if you are working the de-minimis; `cart-vat-nudge` if buying several.

For shortlist mode, emit one block per listing plus a comparison line naming the best landed option across all of them — not the best sticker price.

Out of scope

• Order tracking after purchase. Different surface entirely, not covered.
• Predicting customs behaviour beyond the documented bands in israel-tax.md.
• Local Israeli retailer comparison.
• Changing the selected shipping option in the cart — read-only, per plugin policy.

Validation checklist

1. Ship-to Israel confirmed from the rendered page, not assumed.
2. The panel was expanded — a single option reported from a collapsed row is a failure of this skill, not a listing with one option. Say which you saw.
3. Every option carries a cost and a lead time, or an explicit ? for what the page did not show. No inferred values.
4. FX rate printed with source and date on any USD figure.
5. A recommendation was made and justified in one sentence.
6. If selectors had to be discovered, reference/selectors.md was updated.