Skip to main content
The Effect Coffee Shop MCP server registers two prompts that pre-load live shop data and return a ready-to-send message for an LLM. Clients invoke prompts by name and supply the required parameter; the server fetches current state and interpolates it into the prompt text.

recommend-drink

Suggest a drink from the available menu for a given occasion.
PropertyValue
Namerecommend-drink
Parameteroccasion (string)

Parameters

occasion
string
required
A short description of the context or mood for which a drink should be recommended. The server provides auto-completion with the following suggestions: morning rush, afternoon break, late night, decaf.

Behavior

When invoked, the prompt fetches the current menu from CoffeeOrderApp and constructs the following message for the LLM:
Recommend one drink for "{occasion}" from this menu:
{prettyJson(menu)}
The LLM receives the full MenuItem[] array serialized as pretty-printed JSON, giving it complete information about available drinks, pricing, milk options, and temperature options before making its recommendation.
Pair this prompt with the coffee://menu resource when you want to inspect the raw menu data yourself before or after the recommendation.

summarize-open-orders

Summarize the current open order queue for a specific operational focus area.
PropertyValue
Namesummarize-open-orders
Parameterfocus (string)

Parameters

focus
string
required
The perspective from which to summarize the queue. The server provides auto-completion with the following suggestions: kitchen, pickup, operations.

Behavior

When invoked, the prompt fetches all orders from CoffeeOrderApp, filters out orders with status picked-up or cancelled, and constructs the following message for the LLM:
Summarize the open order queue for {focus}:
{prettyJson(openOrders)}
The LLM receives the filtered CoffeeOrder[] array — only pending, brewing, and ready orders — as pretty-printed JSON. The focus parameter steers the tone of the summary:
  • kitchen — suitable for baristas preparing drinks; emphasizes what needs to be made.
  • pickup — suitable for front-of-house; emphasizes orders ready for handoff.
  • operations — suitable for managers; provides a broader throughput view.
The filtering logic mirrors the coffee://orders/open resource exactly: orders with status picked-up or cancelled are excluded before the prompt is built.

Build docs developers (and LLMs) love