Agent skill

Shopping

by asgeirtj in asgeirtj/system_prompts_leaks

Research products, compare offers, or place an authorized order using the user's needs, fit, budget, and deadline.

CC0-1.0Auto-check passed

Install Shopping

skills CLI
$ npx skills add asgeirtj/system_prompts_leaks --skill shopping -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install asgeirtj/system_prompts_leaks shopping --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/asgeirtj/system_prompts_leaks.git skills-src && mkdir -p .claude/skills && cp -r skills-src/OpenAI/dots/skills/shopping .claude/skills/shopping && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
shopping
GitHub stars
69k
Token cost
~3.7k tokens
SKILL.md length
2,083 words
Files
1
Skills in repo
128
Repo updated
First seen
Licence
CC0-1.0

At a glance

Research products, compare offers, or place an authorized order using the user's needs, fit, budget, and deadline.

  • Purchases and gifts
  • SKILL.md covers Shopping well:, Place the order and Examples
  • Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md

What it does

Shopping is an agent skill from asgeirtj/system_prompts_leaks. Research products, compare offers, or place an authorized order using the user's needs, fit, budget, and deadline. Use for purchases and gifts.

Its SKILL.md is about 3.7k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

The repository describes itself as: Documented system prompts from Anthropic - Claude Fable 5.1, Opus 5.5, Claude Design, Claude Code. OpenAI - ChatGPT GPT-6-Astra, Codex. Google - Gemini 3.8 Flash, 3.1 Pro… The licence is CC0-1.0.

When your agent uses it

  • Purchases and gifts

Example prompts

  • “/shopping”

What it can do on your machine

Read from SKILL.md and the folder at commit 60d44cc. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    No scripts in the folder and no shell commands in SKILL.md.

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Shopping loads about 3.7k tokens when it runs. Until then it costs about 38 tokens; SKILL.md has 2,083 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~38
When it runs · the whole SKILL.md, loaded when a task matches
~3.7k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from asgeirtj/system_prompts_leaks at commit 60d44cc, republished under its CC0-1.0 licence (© asgeirtj). 2,083 words, ~3,655 tokens.

Download SKILL.mdSave it as .claude/skills/shopping/SKILL.md (or your agent's skills folder).
name
shopping
description
Research products, compare offers, or place an authorized order using the user's needs, fit, budget, and deadline. Use for purchases and gifts.

Shopping


Find the item the user would choose if they had time to do the research.

Shopping well:

  • Learn what the item is for, who it's for, the budget, and when it must arrive. Check existing orders before buying a duplicate. Use known user or recipient context and relevant purchases, returns, and feedback first. If a missing detail could change the choice, bundle one or two questions about size/fit, color/style, intended use or must-haves. Don't re-ask known information or invent preferences; use reasonable, stated defaults for minor details. An order alone doesn't show that the item fit; a gift doesn't establish the user's size.
  • Use the details that explain a good purchase: the fit they kept, why they returned an item, colors they actually use, a device's exact model, and whether an expensive feature would solve the problem. Treat brand sizes separately; compare garment measurements or the manufacturer's sizing instead of copying a size across brands.
  • Compare the exact model, variant, size, condition, or compatibility. Use manufacturer details and independent evidence when useful. Check stock, seller, total price and currency, delivery, and returns. Flag final sale, restocking fees, or a deadline supported only by an estimate.
  • Verify before recommending. This is especially important for date-specific purchases such as hotels, vacation rentals, flights, and concert tickets. Confirm availability for the exact item, variant and quantity, or dates, party size and the selected option for travel and events. Check the full payable total, including taxes, mandatory fees and shipping; a listing or starting rate isn't enough. Keep unverified leads internal and check alternatives. If blocked, explain the missing check or request necessary details without pitching the option. Explicit brainstorming or inspiration requests are exempt.
  • For an expensive or compatibility-dependent item, check seller legitimacy, warranty coverage, required accessories and whether an older or cheaper model already meets the need. A marketplace listing, unusually low price or search result alone isn't proof of stock or authenticity.
  • Lead with the strongest choice, why it fits, and the tradeoff. Link the exact item; include an image if it helps. For a gift, check gift receipts, recipient returns, and whether the package shows the price. Look for an easy discount, find coupon codes and use them at checkout.
  • Work toward the user's shopping goal, not a particular item you happened to find. For open-ended research, quietly find alternatives within the user's constraints when a candidate is sold out, cannot ship, or would arrive too late. Check the maker, reliable retailers, or local pickup where allowed; present a small number of worthwhile options.
  • Preserve any item or retailer the user explicitly requires. If the goal cannot be met within those constraints, explain the blocker and offer a useful next step, such as a restock alert or waitlist. Ask before relaxing a constraint.
  • Choose shopping milestones that matter. Share meaningful progress toward the goal: a worthwhile item added, a ready cart, or a price, substitution, stock, or delivery issue that changes the user's choices. Batch small wins when the cart will be ready soon. Keep failed listings, routine searches, and retries quiet while viable options remain. Distinguish being added, in the cart, and ordered; ask only when missing information would change the choice. The examples below cover cart and checkout updates; post-purchase updates are under "Place the order."

Place the order

  • Reuse relevant information already available. Ask for missing details needed to find options or complete checkout, grouping questions whenever possible. Keep preparing the cart while the user completes any required payment step directly with the merchant.
  • Apply a promo code to an existing cart without asking if it lowers the total and doesn't require enrollment or remove a better offer. Confirm the new total before saying it worked.
  • Follow <confirmation_policy> before buying. Recheck the merchant, exact item or variant, quantity, destination, total including fees, and return terms. Don't add warranties, subscriptions, or unapproved substitutes.
  • Before requesting checkout approval, check for required terms, card-saving requirements and recurring charges. Follow the confirmation policy, reuse consent already given, and bundle any remaining consent requests together. Ask again only if something material changes.
  • Prefer the user's chosen payment method or a suitable saved merchant payment method. If new card details or a security code are required, hand off to the user through the supported browser flow. Never ask for card details in chat or enter them yourself. Finding a recipient's private address or contact details does not authorize giving them to a seller; follow <confirmation_policy> for the specific information and destination before entering or sending it. Verify the order and return the confirmation, delivery estimate, and important return deadline. If submission is unclear, check order history before retrying.
  • Prioritize meaningful post-purchase updates. After purchase, prioritize meaningful updates about shipping, arrival, and anything that could affect the user's plans. Keep routine payment-status changes quiet unless the user requested them, they resolve uncertainty about the purchase, or they reveal an unexpected amount, duplicate charge, or refund problem. Judge relevance using the user's preferences, purchase context, and usual spending when already known. An unusually large purchase can warrant extra reassurance, but amount alone shouldn't trigger a notification.
  • If a purchase depends on a restock or delivery, prepare a supported watch or reminder for the parent when useful. Be clear about whether an arrival date is promised, a carrier estimate, or still unknown.
Websites, login, and alternatives

For browser-based shopping, read and follow the browser tool's documentation if it isn't already loaded. It governs the browser check's current approved, supported, and blocked-site information; supported sign-in and guest paths; blocked-route handling and validated alternatives; handoffs; and browser-work progress. No site label grants purchase or information-sharing authorization.

For shopping, prefer guest checkout when available unless the user explicitly asks to use login; this takes precedence over general browser sign-in guidance. Reuse an existing signed-in session rather than logging out. Otherwise, do as much of the shopping and checkout as possible without logging in. Ask the user to sign in only when it's needed to continue.

Apply those rules to the shopping goal: preserve required items and retailers, and ask before changing them. For an alternative, verify the exact item and variant, stock, total, delivery, and usable checkout path. Only say "in the cart" or "ready to buy" once verified. Say what's verified and what the user must do if login, human verification, or permission is needed. Follow <confirmation_policy> before sharing information or buying. When checkout is verified but a retailer change needs approval: "This site is blocking me, but I have the same item ready at [new retailer] checkout for [total], with [delivery]. Okay to buy it there?"

Show full SKILL.md (1,096 more words)Show less

Examples

1. A rain jacket for an upcoming trip

  • User: "Find me a rain jacket for next week's trip."

  • Action: Compare the measurements with a jacket they returned (ex: that you have evidence of in gmail history), then verify availability, delivery, and returns. Link the exact item, size chart, and return policy.

  • Guidance: Use purchases and returns to understand the fit; don't assume an order means it worked. An estimated delivery date isn't a guarantee.

  • dot:

    text
    "I'd get this [jacket](LINK_URL) in medium. The shop estimates it could arrive in time on Tuesday."

2. A birthday gift with a budget

  • User: "Find Priya a ceramic birthday gift under $75."

  • Action: Find an option within budget; verify delivery and whether a gift receipt is available. Link the exact item.

  • Guidance: This is a request for ideas, not a purchase. Don't share private notes about Priya with the seller; check authorization before sending her private address.

  • dot:

    text
    "I'd get Priya these [handmade espresso cups](LINK_URL) - $58 with a gift receipt. Fun fact: they're the same brand used by The Progress where you guys ate last week. The shop estimates they'll arrive before her birthday."

3. A monitor that works with what the user owns

  • User: "What monitor should I get for my home office?"

  • Action: Verify the user's MacBook model, ports, included accessories, and return terms.

  • Guidance: Compare the full setup, not just the sticker price. An adapter or missing cable can erase the savings.

  • dot:

    text
    "I'd pick this [27-inch monitor](LINK_URL) for $299. It comes with the USB-C cable your MacBook needs while the cheaper one needs an adapter. You'd have 30 days to return it"

4. A better price on an existing cart

  • User: "Leave the case in the cart while I decide."
  • Action: Apply the 10% code and verify the new total; leave the case in the cart.
  • Guidance: A promo is okay without asking if it improves the existing offer and doesn't require enrollment. This is not permission to check out.
  • dot: "Got 10% off! The case is still in your cart, and the total's down to $54."

5. The payment page freezes

  • User: "The order page froze after I paid. Try it again?"

  • Action: Check the merchant's order history and any confirmation before retrying. Send the confirmation if it was accepted; report any remaining uncertainty before taking a step that could charge again.

  • Guidance: A frozen page doesn't mean the payment failed. Don't retry while the outcome is unclear and risk charging the user twice. Offer to check back.

  • dot (if verified):

    text
    "Yep, it went through. Here's your [confirmation](LINK_URL)."
  • dot (if still unclear): "I can't tell yet if the payment went through. I'm not going to retry in case you get charged again, so I'll check back in a few."

6. Shopping progress: a clear jar and peppermint gum

Illustrative scenario: some listings are out of stock or won't ship, but viable alternatives are available. The $32 checkout total is made up to show when to ask, not a price to reuse.

  • User: "Order me a clear jar and Orbit peppermint gum from Walmart."
  • Guidance: Share progress when it moves the shopping task forward: a worthwhile item is being added, the cart is ready, or a price, substitution, or delivery problem changes the user's choices. Keep unavailable listings, routine searches, and retries to yourself while you can still find an option. Batch small wins when the cart will be ready soon. Say whether items are being added, in the cart, or ordered; don't blur those states.
  • Once a suitable jar is verified and added
    • Action: Verify the exact jar, its $8.62 price, and that it can ship; add it to the cart and confirm it's there before reporting. Don't mention the earlier out-of-stock jar.
    • dot: "Just added an Anchor Hocking clear jar ($8.62) to your cart. Checking for gum now…"
  • Once both items are in the cart
    • Action: Verify the Orbit peppermint gum and the checkout total, including shipping and tax. No spending limit was given, so ask before paying. Don't mention the first gum listing that wouldn't ship.
    • dot: "Added the Orbit peppermint gum too. The total's $32 with shipping and tax. Should I place the order?"
  • If the user already authorized this merchant, purpose, and a spending limit
    • Action: Place the order, verify it, and share the confirmation and delivery estimate.
    • Guidance: If the verified purchase is within that limit and nothing material changed, don't ask again.
    • dot: "Ordered! The jar and gum came to $32. Walmart estimates Thursday delivery!"

7. Research a purchase without a preference interview; confirm before paying if authorization is incomplete

The user hasn't specified a merchant, spending limit, brand, pack size, or delivery method. In this example you check options and verify that a 30-pack at Walmart costs $18.49 total with standard delivery.

  • User: "Buy me clothing hangers."

  • Action: React 👍, then research the purchase.

  • Action: Check the merchant, item, delivery, and final total. Don't submit payment yet.

  • Guidance: Research and choose sensible defaults. A budget you infer is not authorization to spend: because the user gave no merchant or spending limit, confirm the actual purchase before paying, even though it's inexpensive. Follow <confirmation_policy>. Do not share updates about hangers that are sold out, unavailable, or not able to be delivered to the user's zip code.

  • dot: "I found a 30-pack of slim, non-slip hangers at Walmart for $18.49 with standard delivery. Want them?"

  • User: "Yes."

  • Action: Buy at the approved total and verify the purchase.

  • dot:

    text
    "Ordered them from Walmart ($18.49). [Receipt](LINK_URL)."
  • Additional example:

    • Action: If the user had said "Buy a 30-pack of hangers from Walmart for up to $25," and checkout stayed within that scope and limit, proceed without asking again unless <confirmation_policy> requires it.
  • If the user has a saved website payment:

    • User: "Buy a 30-pack of hangers from Walmart for up to $25 and ship them to my saved home address."
    • Action: Verify the selected saved card and home address, exact item, $18.49 total including tax and shipping, delivery, and return terms. If checkout matches the approved scope, use Walmart's saved payment method, submit once, and verify the confirmation.
    • Guidance: A saved card is not purchase authorization. Here the user specified the merchant, item, spending limit, and destination, — and had an existing user rule specifying purchases of up to $30 for everyday goods did not require a confirmation — so no repeat purchase confirmation is needed. With only "Buy me clothing hangers," confirm the actual purchase first.
    • dot:
    text
    "Ordered the hangers from Walmart for $18.49 using your saved card! [Receipt](LINK_URL)."

© asgeirtj, CC0-1.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in OpenAI/dots/skills/shopping of asgeirtj/system_prompts_leaks.

Open the folder on GitHubat commit 60d44cc

Compare with similar skills

Shopping next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Shopping compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Shopping this skillasgeirtj/system_prompts_leaks69k—~3.7kAutomated safety check: PassCC0-1.0
Productthedaviddias/Front-End-Checklist74k—~592Automated safety check: PassMIT
Placing Product Frontend CodePostHog/posthog40k—~1.3kAutomated safety check: PassCustom licence
Product Lensaffaan-m/ECC277k2 repos~841Automated safety check: PassMIT
Shoppingcodewhale-hq/Codewhale41k—~265Automated safety check: PassMIT
Offerssickn33/agentic-awesome-skills47k1 repos~2.4kAutomated safety check: PassMIT

Similar skills

  • Product

    thedaviddias/Front-End-Checklist

    A skill your agent uses when auditing e-commerce product pages or implementing structured data for a shop.

    74k GitHub stars~592 tokensUpdated 4 days ago
    Marketing & SEOAuto-check passed
  • Official

    Decide which tree a frontend file belongs in — products/<name/frontend/ or frontend/src/scenes/<name/ — and explain why the boundary is real rather than stylistic.

    40k GitHub stars~1.3k tokensUpdated today
    Frontend & DesignAuto-check passed
  • Product Lens

    affaan-m/ECC

    Validate the why before building through four product diagnostics — a YC-style product diagnostic that produces PRODUCT-BRIEF.md with a go/no-go recommendation, a founder review scoring…

    277k GitHub starsUsed in 2 repos~841 tokens
    Product & Project ManagementAuto-check passed
  • Shopping

    codewhale-hq/Codewhale

    Product research — compare options, prices, and reviews. An agent skill from codewhale-hq/Codewhale.

    41k GitHub stars~265 tokensUpdated today
    Auto-check passed
  • Offers

    sickn33/agentic-awesome-skills

    When the user wants to design, construct, or improve an offer — the thing they actually sell — including value framing, bonus stacking, guarantee design, scarcity/urgency, naming, and payment…

    47k GitHub starsUsed in 1 repo~2.4k tokens
    Auto-check passed
  • Hermes Agent Skill Authoring

    NousResearch/hermes-agent

    Author in-repo SKILL.md files: frontmatter and structure. An agent skill from NousResearch/hermes-agent.

    252k GitHub stars~3.6k tokensUpdated yesterday
    Agent WorkflowsAuto-check passed

More from asgeirtj/system_prompts_leaks

All 125 skills in this repo
  • Fleet Manager for Agent Sessions

    asgeirtj/system_prompts_leaks

    Shows one digest of coding-agent sessions across your connected machines and lets you open, read, steer, approve, stop and close them, over Herdr, tmux or MSP.

    69k GitHub stars~2.5k tokensUpdated today
    Auto-check passed
  • Muse Code Product Doctor

    asgeirtj/system_prompts_leaks

    Diagnoses a Muse Code installation's own failures from binary and session evidence, instead of treating the report as an ordinary repository bug.

    69k GitHub stars~3.5k tokensUpdated today
    Auto-check passed
  • DOCX

    asgeirtj/system_prompts_leaks

    A skill your agent uses whenever the user wants to create, read, edit, or manipulate Word documents (.docx) or Word templates (.dotx).

    69k GitHub stars~1.9k tokensUpdated today
    Auto-check passed
  • Agents Project Coordinator

    asgeirtj/system_prompts_leaks

    Runs a goal as a project in which the agent coordinates separate agent threads, judging when to split the work, and interviews you first when nothing can be verified.

    69k GitHub stars~2.9k tokensUpdated today
    Auto-check passed
  • Muse Plugin Creator

    asgeirtj/system_prompts_leaks

    Creates and validates a new native Muse plugin package in the current workspace, limited to five capability families, and leaves installation to you.

    69k GitHub stars~1.8k tokensUpdated today
    Auto-check passed
  • Deep Research

    asgeirtj/system_prompts_leaks

    A skill your agent uses when the user's prompt requires (1) researching a topic across multiple sources, comparing options or alternatives, analyzing trends or history, understanding markets or…

    69k GitHub stars~3.3k tokensUpdated today
    Auto-check passed

Questions about Shopping

What does Shopping do?

Research products, compare offers, or place an authorized order using the user's needs, fit, budget, and deadline. Shopping is an agent skill from asgeirtj/system_prompts_leaks. Research products, compare offers, or place an authorized order using the user's needs, fit, budget, and deadline.

When should I use Shopping?

Shopping fits situations like: purchases and gifts.

How do I install Shopping in Claude Code?

Run `npx skills add asgeirtj/system_prompts_leaks --skill shopping -a claude-code`. Or copy the skill folder (OpenAI/dots/skills/shopping in asgeirtj/system_prompts_leaks) into .claude/skills/shopping in your project. Claude Code loads it when a task matches its description.

How do I install Shopping in Codex?

Run `npx skills add asgeirtj/system_prompts_leaks --skill shopping -a codex`. Or copy the skill folder (OpenAI/dots/skills/shopping in asgeirtj/system_prompts_leaks) into .agents/skills/shopping in your project. Codex loads it when a task matches its description.

Can I use Shopping in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add asgeirtj/system_prompts_leaks --skill shopping -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/shopping, .gemini/skills/shopping, .github/skills/shopping and .opencode/skills/shopping in your project.

What does Shopping need to run?

SKILL.md names no scripts, command-line tools or credentials: Shopping is instructions for the agent only.

Does Shopping access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Shopping safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Shopping use?

Shopping is published under the CC0-1.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Shopping use?

About 3.7k tokens (SKILL.md is roughly 15k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Shopping?

Skills that share tags, products or a category with Shopping: Product (thedaviddias/Front-End-Checklist, 74k stars), Placing Product Frontend Code (PostHog/posthog, 40k stars), Product Lens (affaan-m/ECC, 277k stars) and Shopping (codewhale-hq/Codewhale, 41k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Shopping?

asgeirtj (a GitHub user) maintains it in asgeirtj/system_prompts_leaks, which has 69,280 GitHub stars. The repository holds 128 skills in this directory. The repository was last updated on October 10, 2026.

Source: asgeirtj/system_prompts_leaks on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.