{
  "title": "MCP Tool Schema Designer",
  "description": "Drafts a well-scoped MCP (Model Context Protocol) tool definition: name, description, input schema, and output shape.",
  "category": "Agentic AI and Multi-Agent Systems",
  "tags": [
    "mcp",
    "tools",
    "schema"
  ],
  "templateContent": "You are designing an MCP tool. Capability to expose: {{capability_description}}\nInputs it needs: {{inputs_description}}\nWhat it returns: {{outputs_description}}\n\nProduce:\n1. A tool name (snake_case, verb-first) and a one-sentence description written for a model deciding whether to call it.\n2. A JSON Schema for the input parameters, marking required fields and giving each a description a model can act on.\n3. The output shape (fields and types), including how errors are represented.\n4. One example call and one example response.\n5. A note on what this tool must NOT be used for, if its name/description alone could be misread.",
  "variables": [
    {
      "name": "capability_description",
      "label": "Capability to expose",
      "type": "multiline",
      "required": true,
      "default": null,
      "allowedValues": null
    },
    {
      "name": "inputs_description",
      "label": "Inputs it needs",
      "type": "multiline",
      "required": true,
      "default": null,
      "allowedValues": null
    },
    {
      "name": "outputs_description",
      "label": "What it returns",
      "type": "multiline",
      "required": true,
      "default": null,
      "allowedValues": null
    }
  ],
  "usageInstructions": "Describe the capability in plain language first; let the prompt derive the schema rather than writing JSON Schema by hand and asking for a description to match it.",
  "limitations": "Produces a design, not a working server implementation; you still need to wire it to your actual MCP server code."
}