OpenParable
Skip to content
← Back to skills

Luckin Ordering

Apps & Automation
luckin-ordering

Luckin Ordering is an Agent Skill for apps & automation published in the AIDotNet/OpenCowork repository on GitHub: a SKILL.md instruction file an AI agent can load to do this task.

Order Luckin Coffee for store pickup with the My Coffee extension tools. Use when the user asks to order coffee, buy a Luckin drink, check a coffee order, or cancel one. Covers store selection, product search, SKU customization, order preview, payment QR, and order status.

View on GitHub ↗
QUICK START

How to use this skill

Bring this guide into your coding agent with a prompt tailored to the tool you use.

  1. Open your project in Codex.
  2. Copy and paste the prompt below.
  3. Review files and risks before approving any action.
Prompt to paste
Inspect the Agent Skill at https://github.com/AIDotNet/OpenCowork/blob/HEAD/resources/extensions/my-coffee/skills/luckin-ordering/SKILL.md and all supporting files. Report provenance, permissions, scripts, dependencies, and security risks. Propose an exact installation into .agents/skills/luckin-ordering, but wait for my approval before writing files or running scripts.

Copying this prompt does not install or run the skill. Review third-party files before use.

Luckin Ordering Workflow

Order Luckin Coffee for store pickup using the My Coffee extension tools (extension__my-coffee__*). Follow this workflow as a contract; the gates below exist because the upstream API rejects out-of-order calls.

Preflight

  1. Run check_token. If the token is missing, stop and walk the user through Settings -> Extensions -> My Coffee before anything else.
  2. If tool parameters ever look stale or a call fails with a schema error, run list_tools and re-check the current schema before retrying.

Hard Rules

  • Pickup only. For delivery requests, ask whether store pickup works instead; do not improvise a delivery flow.
  • Never pick a store for the user. Show candidates from query_shop_list and wait for an explicit choice. Only set locationIsPrecise: true when the user gave exact coordinates.
  • Never guess SKUs. Customizations (temperature, sweetness, milk, size) go through query_product_detail then switch_product; search results alone do not identify a SKU.
  • preview_order is a mandatory gate before create_order: same store, same product list. The create call is rejected otherwise. Coupons returned by preview are passed through automatically.
  • Show only the payment QR (payOrderQrCodeUrl) after creating an order. Show the pickup code only when query_order_detail reports the order is paid and returns pickup info.
  • create_order and cancel_order are mutating actions; confirm intent with the user first.

Flow

  1. Ask what drink they want and roughly where they are (or reuse the store from earlier in the conversation).
  2. query_shop_list -> user picks a store.
  3. search_product in that store -> if customization is requested, query_product_detail + switch_product to get the final SKU.
  4. preview_order -> confirm the final price with the user.
  5. create_order -> show the payment QR and tell the user to scan it.
  6. On request, query_order_detail for status/pickup code; cancel_order to cancel.

Keep replies short and warm. Lead with the result (store found, price confirmed, QR ready), not with tool mechanics.