ChatGPT prompt for system design

Design [SYSTEM TO DESIGN] for [USERS AND USE CASES]. The expected scale is [SCALE AND SLOS], and these constraints are non-negotiable: [CONSTRAINTS]. The existing stack, integrations, and deployment environment are [CURRENT CONTEXT]. The main risks or quality priorities are [PRIORITIES].

Treat this as a technical design review. Ask up to 3 clarifying questions only if a required input is missing; otherwise state reasonable assumptions separately. Do not invent vendor guarantees, benchmarks, compliance certifications, existing services, or load-test results.

Produce a design document with:
1. Scope, non-goals, functional requirements, and measurable non-functional requirements.
2. Back-of-the-envelope capacity estimates: requests, storage, bandwidth, and peak versus average load. Show the arithmetic and label assumptions.
3. A component diagram in Mermaid plus a request flow for the most important write and read paths.
4. Data model and ownership boundaries, including primary keys, indexes, retention, and the consistency model required for each critical operation.
5. API or event contracts with example requests, responses, idempotency behavior, pagination or ordering rules, and versioning decisions.
6. Scaling and reliability: caching, partitioning or sharding if justified, queues, retries with backoff, dead-letter handling, rate limits, timeouts, and observability.
7. Security and privacy controls relevant to the data and actors: authentication, authorization, secrets, audit logging, and abuse prevention.
8. Failure modes and trade-offs, including regional dependency, duplicate delivery, stale reads, hot keys, partial writes, and recovery behavior.
9. A phased implementation plan with the smallest viable first release and tests needed before scale-up.

Before answering, verify that every component supports a stated requirement, every asynchronous path explains delivery semantics, and no claimed consistency or availability guarantee conflicts with the proposed architecture.

Act as an interviewer and candidate for a [DURATION]-minute system-design interview on [SYSTEM]. The expected scale is [SCALE]. First ask the highest-value clarifying questions, then provide a candidate-style answer in this order: requirements, estimates, high-level design, data model, critical flows, bottlenecks, and trade-offs. Include a simple Mermaid diagram and explain what I would say while drawing it. Keep the first design practical, then describe two scale-up changes. Check that calculations and component choices match the stated scale.

Review this system-design proposal: [DESIGN DOCUMENT]. Its requirements and constraints are [REQUIREMENTS]. Produce an architecture-review memo with strengths, blocking risks, important unanswered questions, and prioritized recommendations. Inspect data ownership, retry and idempotency behavior, consistency claims, capacity assumptions, authorization boundaries, observability, and operational recovery. Do not redesign components without linking each recommendation to a requirement or failure mode. End with an acceptance checklist and identify any claim that needs a load test or prototype.

Design REST or event APIs for [CAPABILITY] used by [CLIENTS]. The core objects are [OBJECTS], and key operations are [OPERATIONS]. Define resources, endpoints or topics, request and response examples, validation errors, authentication and authorization, idempotency, pagination, ordering, concurrency control, versioning, and webhook delivery semantics if relevant. Explain the data ownership boundary and identify which operations need transactions. Do not assume fields or permissions not supplied. Check that retries cannot create duplicate side effects and that error responses let clients recover safely.