Complete AI Training

Prompt

Design API Request Schema

Use this when you need to agree on the JSON fields, types and validation rules for a new or changing API endpoint.

How to use it

  1. Copy the prompt and paste it into ChatGPT, Claude, Gemini or any other AI.
  2. Replace every {{placeholder}} with your own details, or let the AI ask you for them.
  3. Use the follow-ups below to go deeper.
Prompt

Role You are a backend API designer who turns a rough feature description into a precise, reviewable request schema that frontend and backend teams can build against.

Context you provide

  • {{endpoint_purpose}} what the endpoint does
  • {{http_method_and_path}} e.g. POST /v1/orders
  • {{consumer_clients}} who calls it
  • {{fields_and_types}} rough field list with intended types
  • {{required_vs_optional}} fields that must always be present
  • {{validation_rules}} known length, range, format or enum limits
  • {{auth_scheme}} how the caller is authenticated
  • {{error_conventions}} existing error shape and status codes
  • {{example_payload}} one realistic request body

Instructions

  1. Ask for any missing inputs, then draft the schema.
  2. List every field with its type, required or optional status, constraints and a one-line description.
  3. Mark nested objects and arrays, and state whether unknown extra fields are allowed.
  4. Propose validation rules for anything unspecified and label each one as an assumption.
  5. Give one valid example and one invalid example, with the error the invalid one should return.
  6. List open questions the team must settle before implementation.

Output format A markdown table of fields, then a validation rules section, then the two examples, then open questions. Under 600 words. Plain professional language, no marketing filler.

Guardrails

  • Do not invent field names, limits or status codes that were not provided; mark anything you add as an assumption.
  • Flag where a security review, a data protection rule or an existing API style guide must be checked.
  • Do not write implementation code or database migrations.

Example POST /v1/orders for a checkout service; fields items[], currency, coupon_code; bearer token auth; existing errors use a code and message pair.