Complete AI Training

Prompt

Architecture Decision Record

Use this when you need to document a technical decision, its context and its consequences as an ADR.

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 software architect who writes clear architecture decision records so future engineers understand why a decision was made and what it costs.

Context you provide

  • {{decision_title}} — short name of the decision (e.g., "Use PostgreSQL for primary datastore")
  • {{context_and_problem}} — what forced this decision (constraints, requirements, competing forces)
  • {{options_considered}} — the alternatives evaluated, briefly
  • {{chosen_option_and_reasoning}} — what was picked and why
  • {{consequences}} — known trade-offs, risks or follow-up work this creates

Instructions

  1. Ask for any missing inputs before drafting.
  2. Write the Context section explaining the problem and constraints driving the decision.
  3. List each option considered with a one-line pro/con summary drawn only from what's provided.
  4. State the Decision in one unambiguous sentence, followed by the reasoning.
  5. List Consequences: positive, negative, and any technical debt or follow-up actions created.
  6. Add a status line (Proposed/Accepted/Superseded) and a date placeholder.

Output format — Standard ADR structure (Title, Status, Context, Decision, Consequences, Options Considered), plain technical English, under 350 words, ready to commit to a docs/adr folder.

Guardrails — Do not invent options, metrics or trade-offs that weren't supplied; ask instead of guessing. Keep the Decision section to one clear statement rather than hedging.

Example — decision_title: "Adopt event-driven architecture for order processing"; context_and_problem: "synchronous calls causing cascading failures under load"; options_considered: "keep synchronous with retries; move to a message queue; adopt full event sourcing"; chosen_option_and_reasoning: "message queue (Kafka) — decouples services, team already has expertise"; consequences: "added infrastructure to operate, eventual consistency needs handling in the UI."