planlisted

# AEP Phase 2 — Plan Do not write production code in this phase. The output is a selected approach and a short spec. ## 1. Brainstorming loop: generate → critique → refine **Generate.** Draft 2–3 genuinely distinct approaches (different architecture or mechanism — not the same idea with cosmetic variations). **Critique.** Attack each approach skeptically against this matrix; one line per cell, no diplomacy: | Approach | Failure modes | Hidden coupling | Scale limits | Migration/rollback | Maintenance cost | |---|---|---|---|---|---| Add a one-line **pre-mortem** per approach: "It is six months later and this approach failed — the most likely reason is ___." **Refine or discard.** Fix what the critique exposed, or drop the approach. If no approach survives, loop again with what the critique taught you — do not proceed with a plan you just demonstrated is broken. **Select.** Pick one and record *why it beat the others* in 1–2 sentences. That rationale goes into the spec and later into §P.4 (Decision Log) if the decision is durable. ## 2. Tech spec (short, mandatory) ``` ## Spec: <task> Approach: <selected approach + why it won> Files: <files to create/modify> Data/API: <schema, contract, or endpoint changes; "none" if none> Migration: <forward path + rollback path; "n/a" if none> Test plan: <which tests prove which gaps closed — map to Gap Analysis numbers> Acceptance: <numbered, independently checkable criteria — the contract Phase 4 walks one by one> Risk