Data as of

HowToCook-MCP Server for Cursor: Setup & Config

Cursor: HowToCook-MCP brings the HowToCook recipe collection to AI assistants.

What you can do with HowToCook-MCP in Cursor

Useful when an assistant needs recipe details and restriction-aware weekly meal planning together.

  • Plan weekly meals around allergies and disliked ingredients.
  • Choose a meal combination for a group of one to ten people.

Before you start: HowToCook-MCP in Cursor

  • Node.js 16.0.0 or later
  • npm or yarn
  • An MCP-compatible client

For the Cursor npm configuration, the README instructs users to run npm i -g howtocook-mcp first to avoid client creation failures. Source builds and HTTP/SSE alternatives are also described. No required environment variables are stated.

HowToCook-MCP Cursor configuration

  1. Choose Cursor configuration scope

    For HowToCook-MCP Server, open Project: .cursor/mcp.json User: ~/.cursor/mcp.json. Merge into mcpServers at project or user scope.

  2. Add HowToCook-MCP Server to Cursor

    Use the HowToCook-mcp entry below in mcpServers. Keep HowToCook-MCP Server's documented stdio connection fields and your other Cursor entries.

  3. Check HowToCook-MCP Server in Cursor

    HowToCook-MCP Server: Enable in Cursor’s MCP settings; inspect tools.

Config file · Project: .cursor/mcp.json

User: ~/.cursor/mcp.json

JSON
{
  "mcpServers": {
    "HowToCook-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "howtocook-mcp"
      ],
      "env": {}
    }
  }
}

HowToCook-MCP tools available to Cursor

HowToCook-MCP's stored tool list documents 5 tools; compare it with Cursor's connected tool list, because the stored list is not a live check.

mcp_howtocook_getAllRecipes

Find mcp_howtocook_getAllRecipes under HowToCook-MCP in Cursor's MCP settings; inspect its arguments when the agent requests a call.

Return the full recipe collection; the README warns of large context usage.

mcp_howtocook_getRecipesByCategory

Find mcp_howtocook_getRecipesByCategory under HowToCook-MCP in Cursor's MCP settings; inspect its arguments when the agent requests a call.

mcp_howtocook_recommendMeals

Find mcp_howtocook_recommendMeals under HowToCook-MCP in Cursor's MCP settings; inspect its arguments when the agent requests a call.

mcp_howtocook_whatToEat

Find mcp_howtocook_whatToEat under HowToCook-MCP in Cursor's MCP settings; inspect its arguments when the agent requests a call.

Recommend a dish combination for the specified group size.

mcp_howtocook_getRecipeById

Find mcp_howtocook_getRecipeById under HowToCook-MCP in Cursor's MCP settings; inspect its arguments when the agent requests a call.

Look up a recipe by name or identifier, including ingredients and cooking steps.

HowToCook-MCP: full stored tool descriptions

Troubleshooting HowToCook-MCP in Cursor

  • Cursor must find npx howtocook-mcp (Node.js/npm) for HowToCook-MCP Server. Enable the project/user entry in MCP settings and inspect its tools.
  • HowToCook-MCP Server uses stdio here: Cursor must keep the configured process running; protocol output belongs on stdout and diagnostics on stderr.
  • Known limitation: Retrieving every recipe can use excessive assistant context.
  • Known limitation: The documented recommendation range is limited to ten diners.
  • General Cursor troubleshooting lives in the /clients/cursor guide.

HowToCook-MCP Server — details. Cursor guide

Frequently asked questions

Cursor's HowToCook-MCP Server entry belongs under mcpServers in .cursor/mcp.json (project) or ~/.cursor/mcp.json (user). Confirm the active scope in MCP settings.

Cursor's HowToCook-mcp example uses stdio.

HowToCook-MCP Server's Cursor configuration invokes npx howtocook-mcp (Node.js/npm).

HowToCook-MCP Server's stored list contains 5 tools, including mcp_howtocook_getAllRecipes, mcp_howtocook_getRecipesByCategory, mcp_howtocook_recommendMeals. Compare those names with Cursor's connected tool list; the stored list is not a live check.

The documented peopleCount parameter accepts values from one to ten.

That option exists, but the README cautions that it can create a very large context payload.