AI & LLM

Prompt Template (Reusable) JSON Example

A JSON example of a reusable LLM prompt template — includes the template string with variables, input schema, model parameters, and few-shot examples. Copy-ready for prompt management.

JSON

Generate mock data Open in Formatter

{
  "id": "prompt_classify_ticket_v3",
  "name": "Classify support ticket",
  "version": 3,
  "model": "gpt-4o",
  "parameters": {
    "temperature": 0,
    "max_tokens": 200
  },
  "template": "Classify the support ticket into one of: billing, technical, account, other.\n\nTicket: {{ticket_text}}\n\nRespond with only the category.",
  "variables": [
    {
      "name": "ticket_text",
      "type": "string",
      "required": true,
      "description": "The raw customer ticket message"
    }
  ],
  "tags": [
    "support",
    "classification"
  ],
  "updatedAt": "2025-05-15T09:00:00Z"
}

Field Reference

idrequiredstringStable identifier used to fetch the template at runtime

versionrequiredintegerIncremented on each edit so you can pin and roll back prompts

templaterequiredstringThe prompt body with {{variable}} placeholders to interpolate

variablesrequiredarray<object>Declared inputs with types — lets you validate before rendering

parametersoptionalobjectDefault model settings (temperature, max_tokens) for this prompt

Variants

Few-shot templateIncludes example input/output pairs to steer the model's format.

Few-shot template

Generate mock data Open in Formatter

{
  "id": "prompt_extract_entities_v1",
  "template": "Extract the person and company from the sentence.",
  "examples": [
    {
      "input": "Ravi joined Acme in 2021.",
      "output": {
        "person": "Ravi",
        "company": "Acme"
      }
    },
    {
      "input": "Mia leads design at Globex.",
      "output": {
        "person": "Mia",
        "company": "Globex"
      }
    }
  ],
  "model": "gpt-4o",
  "parameters": {
    "temperature": 0
  }
}

Common Use Cases

→Versioning and storing prompts outside application code
→A/B testing prompt variations and rolling back bad changes
→Sharing reusable, parameterised prompts across a team

promptprompt templateprompt engineeringfew-shotllmprompt management

Validate or format this JSON

One click loads this exact example into the tool — no copy-paste needed. Format it, validate it, explore the tree, or generate TypeScript types instantly.

Format Validate Explore tree Generate TypeScript Minify

Frequently Asked Questions

Externalising prompts into versioned JSON lets non-engineers edit them, enables A/B testing and rollback, and decouples prompt changes from deployments. You fetch the template by id and version at runtime and interpolate the variables.

Use a clear placeholder syntax such as {{variable}} and declare each variable with a name, type, and whether it's required. Validate and escape inputs before interpolation to avoid prompt injection and malformed prompts.

Related JSON Examples

AI & LLMChat Conversation HistoryOrdered messages array with system, user, and assistant roles for LLM chats.AI & LLMLLM Structured Output (JSON Schema)Structured output request with a JSON Schema and the model's schema-valid response.AI & LLMFine-Tuning Dataset Example (JSONL)A single fine-tuning training example in the chat messages JSONL format.