browse-add-items-guest▌
instacart.com/browse-add-items-guest-gf4i37 · updated May 21, 2026
MDX-style export adds YAML metadata + attribution linking explainx.ai and this canonical listing URL.
Search Instacart for products and add them to the cart as a guest (no login). Routes around the un-closable email-capture modal that intercepts user-cursor clicks after the first add-to-cart action.
| name | browse-add-items-guest |
| title | Instacart Search and Add to Cart (Guest) |
| description | >- Search Instacart for products and add them to the cart as a guest (no login). Routes around the un-closable email-capture modal that intercepts user-cursor clicks after the first add-to-cart action. |
| website | instacart.com |
| category | grocery |
| tags | - grocery - instacart - cart - guest-checkout - modal-dismissal - anti-bot |
| source | 'browserbase: agent-runtime 2026-05-19' |
| updated | '2026-05-19' |
| recommended_method | browser |
| alternative_methods | - method: browser rationale: >- Instacart exposes no public guest-cart API. The site is a JS-heavy React app behind an Akamai/PerimeterX-style edge. All add-to-cart state lives in the front-end session and is committed via XHR/GraphQL calls that are tightly coupled to fingerprinted cookies. A residential-proxy + verified Browserbase session is the only reliable surface. - method: api rationale: >- Not viable for guest carts — verified across iteration: there is no documented public endpoint, and the internal GraphQL/Stripe-fingerprinted XHRs require auth + cookies + bot-detection signatures that fail to reproduce out-of-band. |
| verified | true |
| proxies | true |
Instacart Search and Add to Cart (Guest)
Purpose
Search Instacart.com for grocery items and add them to a guest cart (no Instacart account / no login). The skill returns the resulting cart contents — item names, sizes, per-item prices, and the cart subtotal — together with the storefront URL the items were added from. The skill is read/write (it mutates a session-scoped cart) but it never logs in, never submits payment, and never completes a checkout. The cart lives entirely in the Browserbase session cookies and is discarded when the session ends.
When to Use
- An agent collecting a grocery list and pricing it in real time across local retailers without a user account.
- A meal-planner or recipe-cost assistant that needs
current_price × quantitytotals for a basket. - Price-comparison flows that pre-build a hypothetical cart on Instacart and then hand the user a "Continue in Instacart" deep-link.
- Anywhere a guest, ephemeral, no-PII cart is the desired surface — checkout / address / payment / membership flows are out of scope and belong in separate skills.
Workflow
Mandatory session config:
browse cloud sessions create --keep-alive --proxies --verified. A bare session is fingerprinted and the search results page renders an empty retailer list.--proxies(residential) is required to get realistic retailer coverage;--verified(advanced stealth) is required to avoid the page-load fingerprint probe escalating to a reCAPTCHA challenge (the page already includes invisible reCAPTCHA —k=6LeN0vMZAAAAAIKVl68OAJQy3zl8mZ0ESbkeEk1m— that activates on stealth failure).
1. Open a search surface
Three URL shapes work. Pick by intent:
| Intent | URL |
|---|---|
| Cross-retailer search across every nearby store | https://www.instacart.com/store/s?k={query} |
| Browse one retailer's storefront (deals, aisles) | https://www.instacart.com/store/{retailer}/storefront |
| Search within one retailer | https://www.instacart.com/store/{retailer}/s?k={query} |
{retailer} is the slug from the storefront URL (e.g. safeway, costco, kroger, 7-eleven, grocery-outlet). The cross-retailer search auto-groups results into per-retailer carousels.
export BROWSE_SESSION="$sid"
browse open "https://www.instacart.com/store/s?k=milk" --remote
browse wait timeout 3500 --remote # initial render is async; ~2.5–3s for the retailer carousels
The page-load process auto-renders without the auth modal — the modal does not appear until the first add-to-cart action.
2. First add — direct snapshot click is fine
The very first add-to-cart action (before the auth modal has ever appeared) works with the normal click pattern:
browse snapshot --remote
# Find a ref like: [1-4947] button: Add 1 ct Lucerne Whole Milk
browse click "[1-4947]" --remote
browse wait timeout 2500 --remote
Two side effects occur:
- The cart counter in the header increments (e.g. "View Cart. Items in cart: 1" → "Add $X to get $0 delivery fee 1").
- The page navigates to
/store/{retailer}/storefrontAND the un-closable "$0 delivery fee on your first 3 orders" auth modal appears as a<div role="dialog">overlay. Don't try to close it. See gotchas — there is no close button and it cannot be dismissed by Escape, backdrop click, or any visible UI affordance.
3. Subsequent adds — use JS .click() to bypass the modal
After the modal first appears, browse click (native mouse-event) is blocked by the dialog's overlay even when targeting buttons outside the modal. The fix: dispatch the click event directly on the target button via browse eval. The underlying React handlers fire normally — the modal blocks the mouse but not the synthetic event.
# Add by aria-label — works whether the modal is visible or not.
browse eval '(() => {
const b = document.querySelector("button[aria-label=\"Add 1 ct Lucerne Whole Milk\"]");
if (!b) return "not-found";
b.click();
return "clicked";
})()' --remote
browse wait timeout 2000 --remote
Each Add 1 ct {name} aria-label is unique per product on the page. After the click succeeds the button morphs into <button aria-label="Decrement quantity of {name}"> plus a + button labeled Increment quantity of {name} — that DOM change is the success signal.
To enumerate addable products from the live page (works even with the modal up — document.querySelector is not modal-blocked):
browse eval '(() => Array.from(document.querySelectorAll("button[aria-label^=\"Add 1 ct\"]")).slice(0,20).map(b => b.getAttribute("aria-label")))()' --remote
4. Reading the cart
The cart counter button lives in the header. Its text shifts based on state: View Cart. Items in cart: 0 (empty) or Add $X.XX to get $0 delivery fee N (non-empty, where N is the item count). To open the cart drawer:
browse eval '(() => {
const cart = Array.from(document.querySelectorAll("button"))
.filter(b => /delivery fee|View Cart/i.test(b.textContent || ""))[0];
cart?.click();
})()' --remote
browse wait timeout 2000 --remote
The drawer renders as another <div role="dialog"> titled "Personal {Retailer} Cart, Shopping in {ZIP}", with each line item showing name, size, current price (and any strikethrough original), and per-line + / 1 ct / trash controls. The drawer also surfaces:
Item subtotal(with discounts already applied)$X Min. to checkout(typically $10 — guest carts under the minimum can still be assembled but not checked out)- "Add $X to get $0 delivery fee" progress
Extract cart contents with the snippet below — it's resilient to the modal being layered on top of the drawer:
browse eval '(() => {
const drawers = Array.from(document.querySelectorAll("[role=dialog]"))
.filter(d => /Personal .* Cart/i.test(d.textContent || ""));
if (drawers.length === 0) return { error: "cart-drawer-not-open" };
const drawer = drawers[0];
const items = Array.from(drawer.querySelectorAll("li, [class*=cart-item], [class*=CartItem]"))
.map(li => (li.textContent || "").replace(/\s+/g, " ").trim())
.filter(t => t.length > 5 && /\$/.test(t));
const subtotalMatch = (drawer.textContent || "").match(/Item subtotal[^$]*\$([0-9.]+)/);
return JSON.stringify({
item_lines: items.slice(0, 50),
subtotal: subtotalMatch ? "$" + subtotalMatch[1] : null
});
})()' --remote
5. (Optional) Make the page snapshottable / browseable visually
For workflows that need browse snapshot to return a usable accessibility tree (e.g. for navigating aisles, browsing categories, or scrolling through search results visually), neutralize the modal without dismissing it as the user would — Instacart never lets you. Three DOM mutations are required together; doing only one or two leaves the page inert:
browse eval '(() => {
// 1) Remove the auth dialog itself
let removed = 0;
document.querySelectorAll("[role=dialog]").forEach(d => {
const t = d.textContent || "";
if (t.includes("delivery fee on your first 3 orders") ||
(t.includes("Or continue with") && t.includes("Continue"))) {
d.remove();
removed++;
}
});
// 2) Remove the modal-open body class that re-applies overflow:hidden and pointer-events:none
document.body.classList.remove("body--auth-modal-open");
// 3) Clear aria-hidden on the app shell — Instacart sets aria-hidden="true" on
// div#js-app while the modal is open, which blanks the accessibility tree.
document.getElementById("js-app")?.removeAttribute("aria-hidden");
document.querySelectorAll("[aria-hidden=\"true\"]").forEach(el => {
if (el.querySelectorAll("button, a, input").length > 3) el.removeAttribute("aria-hidden");
});
return removed;
})()' --remote
After this, browse snapshot returns the full tree and browse click [ref] works on visible buttons. The auth-modal-dismissal effect is per page load — it must be re-applied after every navigation. The modal re-mounts on every route change while the session remains unauthenticated.
For programmatic add-to-cart flows that don't need a clean snapshot, you can skip step 5 entirely and just JS-click everything.
6. Release the session
browse cloud sessions update "$sid" --status REQUEST_RELEASE
The guest cart is not persisted — closing the Browserbase session discards it. If the cart contents are the deliverable, extract them in step 4 before releasing.
Site-Specific Gotchas
-
The "$0 delivery fee on your first 3 orders" modal has no Close affordance. The DOM contains two
<button aria-label="Close">elements inside the dialog, but both are rendered withwidth:0; height:0(verified viagetBoundingClientRect()returning all-zeros andoffsetParent === null). There is no visible X. Escape key does not dismiss. Clicking the modal backdrop does not dismiss. The only "exits" the modal offers are paths into authentication (email, Google, Phone, Log in). Dismissal requires DOM mutation (d.remove()); see step 5. -
Snapshot returns empty unless aria-hidden is cleared. When the modal is open, Instacart sets
aria-hidden="true"on<div id="js-app">, which contains 60+ buttons.browse snapshothonors aria-hidden and returns a single empty StaticText. Symptom: snapshot looks like the page is blank even though the screenshot shows full content. Fix is in step 5 —document.getElementById("js-app").removeAttribute("aria-hidden")plus removingbody--auth-modal-openclass. -
Native
browse clickis intercepted by the modal even on buttons outside the modal. When the auth dialog is open, the modal's overlay sits above the page and absorbs cursor events. Symptom:browse click [ref]returns{"clicked": true}but the page state doesn't change (cart counter doesn't increment, drawer doesn't open). Usebrowse evalwithbutton.click()instead — the synthetic event reaches the React handler directly and the modal's pointer-events shield is irrelevant. This is the single most important pattern in the skill. -
First add is special. The auth modal does not mount until the first successful add-to-cart. So the very first
Add 1 ctclick works fine withbrowse click [ref]. From the second add onward — whether on the same page or any subsequent page — assume the modal is up and use JS-click. -
The cart counter button doubles as the cart drawer trigger. It has two text variants:
View Cart. Items in cart: N(empty/initial) andAdd $X.XX to get $0 delivery fee N(with items). Selecting by aria-label or stable class is fragile — match by regex ontextContentagainst/delivery fee|View Cart/i. -
Cart drawer is a separate
<div role="dialog">. When opened with items in cart, there are now two dialogs in the DOM: the auth modal (centered) and the cart drawer (right-rail). Filter by text content (/Personal .* Cart/ifor the cart drawer,/delivery fee on your first 3 orders/for the auth modal) — don't index by position. -
IP-based ZIP geolocation, no URL override. The page picks a ZIP based on the request IP (in our trace,
97818Heppner OR from a Browserbase residential proxy in the Pacific NW). Appending?zip_code=10001to the URL does not override — the page silently ignores the param and continues using the IP-derived ZIP. To change the address you must either (a) click the ZIP button in the header and use the picker UI, or (b) choose a Browserbase proxy region whose egress IP geolocates to the target metro. The retailer set returned by the search is gated on the active ZIP, so this matters: a 97818 search for "milk" surfaces Safeway / Grocery Outlet / CHEF'STORE / 7-Eleven and excludes urban chains like Whole Foods / Target. -
Clicking "Add" on a cross-retailer search result navigates to that retailer's storefront.
/store/s?k=milkshows item carousels grouped by retailer. Clicking[1-XXXX] Add 1 ct Lucerne Whole Milkfrom the Safeway carousel does two things atomically: adds to cart ANDpushState's to/store/safeway/storefront. The cross-retailer search is not a place you stay — it's a routing surface. -
In-store search canonicalizes URL form.
https://www.instacart.com/store/{retailer}/search/{query}returns 404 ("Page not found"). The correct form ishttps://www.instacart.com/store/{retailer}/s?k={query}— note thes(singular) and the?k=query param. -
No public guest-cart API. Internal mutations go through Stripe-fingerprinted XHRs (Stripe
m-outer-*.htmlis injected on every page) and Akamai/PerimeterX-checked GraphQL endpoints. Out-of-band reproduction of the cart-mutation calls fails — the cookies and__shared_params__are bound to the Browserbase session. The browser surface is the only reliable path; treat any "scrape the JSON API" suggestion as a dead end. -
Invisible reCAPTCHA is present on every page. Site key
6LeN0vMZAAAAAIKVl68OAJQy3zl8mZ0ESbkeEk1m. It does not challenge in a verified+proxies session, but a bare-session run was observed to escalate. If your run starts failing with "session not interactive" or repeated XHR 403s, the captcha bframe has likely activated — restart with stealth on. -
Cart minimum is $10 to advance to checkout. Sub-$10 guest carts are valid (the drawer renders correctly, items can be added/removed) but the "Complete your cart" CTA is greyed. This skill stops at the assembled-cart stage; the minimum constraint matters only if downstream steps want to proceed to checkout.
Expected Output
{
"success": true,
"retailer": "Safeway",
"zip": "97818",
"storefront_url": "https://www.instacart.com/store/safeway/storefront",
"items": [
{
"name": "Lucerne Whole Milk",
"size": "128 fl oz",
"quantity": 1,
"unit_price_usd": 3.99
},
{
"name": "Signature SELECT Pie, Blackberry, Lightly Glazed",
"size": "4 oz",
"quantity": 1,
"unit_price_usd": 0.58,
"original_price_usd": 2.30,
"deal": "75% off"
},
{
"name": "Signature SELECT Pie, Lemon, Lightly Glazed",
"size": "4 oz",
"quantity": 1,
"unit_price_usd": 0.58,
"original_price_usd": 2.30,
"deal": "75% off"
},
{
"name": "Oven Joy White Enriched Bread",
"size": "20 oz",
"quantity": 1,
"unit_price_usd": 2.14
}
],
"subtotal_usd": 5.15,
"subtotal_original_usd": 8.59,
"checkout_minimum_usd": 10.00,
"checkout_ready": false,
"ready_delta_usd": 4.85
}
Alternative outcome shapes:
// Guest-cart construction was successful but ZIP didn't yield the requested retailer
{ "success": true, "retailer": "Safeway", "fallback_from_requested": "Whole Foods", "items": [...] }
// Modal-dismissal mutation triggered but item add still failed (rare — usually means
// a captcha challenge surfaced; recommend restart with fresh verified+proxies session)
{ "success": false, "reason": "add_to_cart_silent_failure", "attempts": 3, "cart_count_after": 0 }
// No retailer available for the IP-derived ZIP
{ "success": false, "reason": "no_retailers_for_zip", "zip": "97818", "query": "specialty-vegan-cheese" }
How to use browse-add-items-guest on Cursor
AI-first code editor with Composer
Prerequisites
Before installing skills in Cursor, ensure your development environment meets these requirements:
- ›Cursor installed and configured on your development machine
- ›Node.js version 16.0+ with npm package manager (verify with
node --version) - ›Active project directory or workspace where you want to add browse-add-items-guest
Execute installation command
Execute the skills CLI command in your project's root directory to begin installation:
The skills CLI fetches browse-add-items-guest from GitHub repository instacart.com/browse-add-items-guest-gf4i37 and configures it for Cursor.
Select Cursor when prompted
The CLI will show a list of available agents. Use arrow keys to navigate and space to select Cursor:
Verify installation
Confirm successful installation by checking the skill directory location:
Reload or restart Cursor to activate browse-add-items-guest. Access the skill through slash commands (e.g., /browse-add-items-guest) or your agent's skill management interface.
Security & Verification Notice
We perform automated surface-level scans (Gen AI Scanner, Socket, Snyk) during installation. These checks detect common vulnerabilities but do not guarantee complete security. Always review skill source code and verify the publisher's reputation before production use.
Skills execute code in your development environment. Always verify the publisher's identity, review recent commits, and test in isolated environments before production deployment.
List & Monetize Your Skill
Submit your Claude Code skill and start earning
Use Cases▌
Task Automation & Efficiency
Automate repetitive workflows and reduce manual effort
Example
Generate reports, summarize documents, draft communications
Save 3-5 hours per week on routine tasks
Knowledge Enhancement
Learn new skills, understand complex topics, get expert guidance
Example
Explain concepts, provide examples, suggest learning resources
Accelerate learning and skill development by 2x
Quality Improvement
Enhance output quality through reviews, suggestions, and refinements
Example
Review drafts, suggest improvements, catch errors
Improve work quality by 30-40% with less effort
Implementation Guide▌
Prerequisites
- ›Claude Desktop or compatible AI client with skill support
- ›Clear understanding of task or problem to solve
- ›Willingness to iterate and refine outputs
Time Estimate
15-45 minutes depending on use case complexity
Installation Steps
- 1.Install skill using provided installation command
- 2.Test with simple use case relevant to your work
- 3.Evaluate output quality and relevance
- 4.Iterate on prompts to improve results
- 5.Integrate into regular workflow if valuable
Common Pitfalls
- ⚠Expecting perfect results without iteration
- ⚠Not providing enough context in prompts
- ⚠Using skill for tasks outside its intended scope
- ⚠Accepting outputs without review and validation
Best Practices▌
✓ Do
- +Start with clear, specific prompts
- +Provide relevant context and constraints
- +Review and refine all outputs before using
- +Iterate to improve output quality
- +Document successful prompt patterns
✗ Don't
- −Don't use without understanding skill limitations
- −Don't skip validation of outputs
- −Don't share sensitive information in prompts
- −Don't expect skill to replace human judgment
💡 Pro Tips
- ★Be specific about desired format and style
- ★Ask for multiple options to choose from
- ★Request explanations to understand reasoning
- ★Combine AI efficiency with human expertise
When to Use This▌
✓ Use When
Use when skill capabilities match your task, clear ROI on time saved, and you can validate outputs. Best for repetitive tasks, learning, and quality improvement.
✗ Avoid When
Avoid when task requires deep expertise you can't validate, involves sensitive decisions, or when learning process is more valuable than speed of completion.
Learning Path▌
- 1Familiarize yourself with skill capabilities and limitations
- 2Start with low-risk, non-critical tasks
- 3Progress to more complex and valuable use cases
- 4Build expertise through regular use and experimentation
Discussion
Product Hunt–style comments (not star reviews)- No comments yet — start the thread.
Ratings
4.6★★★★★41 reviews- ★★★★★Pratham Ware· Dec 16, 2024
browse-add-items-guest has been reliable in day-to-day use. Documentation quality is above average for community skills.
- ★★★★★Aanya Patel· Dec 16, 2024
Registry listing for browse-add-items-guest matched our evaluation — installs cleanly and behaves as described in the markdown.
- ★★★★★Lucas Wang· Dec 12, 2024
browse-add-items-guest fits our agent workflows well — practical, well scoped, and easy to wire into existing repos.
- ★★★★★Isabella Sanchez· Dec 4, 2024
browse-add-items-guest reduced setup friction for our internal harness; good balance of opinion and flexibility.
- ★★★★★Anika Lopez· Nov 15, 2024
I recommend browse-add-items-guest for anyone iterating fast on agent tooling; clear intent and a small, reviewable surface area.
- ★★★★★Sakshi Patil· Nov 7, 2024
Solid pick for teams standardizing on skills: browse-add-items-guest is focused, and the summary matches what you get after install.
- ★★★★★Yuki Bhatia· Nov 7, 2024
browse-add-items-guest fits our agent workflows well — practical, well scoped, and easy to wire into existing repos.
- ★★★★★Tariq Jackson· Nov 3, 2024
Registry listing for browse-add-items-guest matched our evaluation — installs cleanly and behaves as described in the markdown.
- ★★★★★Chaitanya Patil· Oct 26, 2024
We added browse-add-items-guest from the explainx registry; install was straightforward and the SKILL.md answered most questions upfront.
- ★★★★★Yusuf Menon· Oct 26, 2024
browse-add-items-guest is among the better-maintained entries we tried; worth keeping pinned for repeat workflows.
showing 1-10 of 41