Structured Output Format Generator
Design a precise output format spec that makes AI responses consistent, parseable, and ready to use.
Document a technical decision as a clean, durable ADR — so engineers joining 18 months from now understand what was decided, why, and what was rejected.
You are a staff engineer who writes Architecture Decision Records that actually get read. Help me document a technical decision as a complete ADR. Decision I'm documenting: [describe the decision — e.g. "choosing between PostgreSQL and MongoDB for our main datastore"] Context: - Problem we're solving: [describe] - System constraints: [performance, scale, team expertise, cost, timeline] - Options I considered: [list the alternatives] - Option I chose: [state it] - Why I rejected the others: [one sentence per alternative] Return a complete ADR: 1. **Title** — [ADR-###] Short Decision Name 2. **Date** — [today] 3. **Status** — Proposed / Accepted / Deprecated / Superseded 4. **Context** — the forces at play that made this decision necessary 5. **Decision** — what was chosen, with clear rationale 6. **Options Considered** — table comparing alternatives across the dimensions that mattered 7. **Consequences** — what becomes easier; what becomes harder; what we're closing off 8. **Revisit Conditions** — what would need to change for this decision to be worth revisiting 9. **Follow-up Actions** — what happens next Write it as if a senior engineer joins the team in 18 months with no context. No jargon without explanation. No decisions without reasoning.
Tech leads, architects, and engineering managers documenting decisions for long-lived codebases.
A complete ADR with title, date, status, context, decision with rationale, options comparison table, consequences, revisit conditions, and follow-up actions — written for a cold reader with no prior context.
Sign in to leave a comment.
No comments yet.
Be the first to share your thoughts.
Works best with
Claude Opus 4
Design a precise output format spec that makes AI responses consistent, parseable, and ready to use.
Breaks down any code snippet into plain English — from high-level intent down to line-by-line mechanics.
Walks through a system design problem the way a staff engineer would — with trade-off analysis, component breakdown, and scaling considerati…