# Devil's Advocate Prioritization Memo

## Position
Do not start with "clean architecture" refactors yet. First, remove business-risk mismatches that can produce incorrect behavior despite clean code.

## What Should Be Prioritized in This Conversation
1. Decide the product contracts that are currently ambiguous.
   - Theme creation rules
   - Schedule GET semantics
   - Canonical defaults
   - Theme scoping for import/export/preferred
2. Approve a short "correctness-first" implementation batch.
   - Theme-scoped source import/export/preferred fixes
   - Theme creation contract alignment
   - Theme ownership validation on source writes
3. Lock a test gating policy before refactoring.
   - Add deterministic tests for scheduler and SSE progress first
   - Avoid relying on live external-provider E2E for CI confidence

## What To Defer (For Now)
1. Large structural decomposition of `synthesis.rs` and major frontend pages.
2. Broad style/lint cleanups.
3. Performance tuning beyond obvious correctness-adjacent fixes.

## Why This Order Is Safer
1. It prevents regressions in core multi-theme behavior before code motion.
2. It creates a stable contract baseline for any future refactor team.
3. It avoids spending effort simplifying code that may be simplified in the wrong direction if contracts change.

## Concrete Next Decision Request
Please confirm these five items so implementation teams can proceed without churn:
1. Draft theme creation allowed or not
2. Schedule GET absent behavior (`404` vs `200 null`)
3. Defaults (`4/3` vs `5/2`)
4. Theme-scoped import/export/preferred policy (yes/no)
5. Fallback label and source-link extraction mode policy