Code Refactoring Wizard
Orchestrate large-scale refactoring with change tracking. Uses code-wizard for discovery.
Context Files
For structure and conventions, read from Docs/context/:
Docs/context/conventions.md- Architecture rules, refactor guidelinesDocs/context/repo-structure.md- Where to place filesDocs/context/packages-map.md- Package boundaries
Changelog Location
All significant changes logged to: Docs/ai/CHANGELOG.md
This file is monitored by docs-updater skill for documentation sync.
Change Categories
| Category | Tag | Example |
|----------|-----|---------|
| Schema | [SCHEMA] | New table, column rename, migration |
| Structure | [STRUCTURE] | File/folder move, directory reorganization |
| API | [API] | RPC signature change, Edge Function update |
| Breaking | [BREAKING] | Removed feature, renamed export |
| Component | [COMPONENT] | React component split/merge |
| Dependency | [DEPS] | Package upgrade, new import |
Workflow
1. Discover (use code-wizard)
/code-wizard "find all usages of ai_plan_quotas table"
/code-wizard "where is subscription logic implemented"
2. Plan
- Current state → Target state
- Migration path
- Rollback strategy
- Affected files list
3. Execute
Order: Database → Backend → Frontend → Tests
4. Log to CHANGELOG.md
## 2026-01-08 - Subscription System Refactor
### [SCHEMA] Replace per-feature quotas with token pools
**Before:**
- `ai_plan_quotas` table (40+ rows)
- Per-feature token limits
**After:**
- `subscription_plans` table
- `token_pools` table
- `ai_operations` table
**Impact:**
- Files: 15 | Migration: Yes | Breaking: Yes
**Related:** #20260108_token_pool_system
5. Sync Docs
/docs-updater "sync changelog to API docs"
Query Changes
To answer "what changed in X system":
- Read
Docs/ai/CHANGELOG.md - Filter by date/category/system
- Summarize
Example response:
## Subscription System Changes (Jan 2026)
1. [SCHEMA] Token pool migration (Jan 8)
- Replaced ai_plan_quotas → unified token system
- 15 files, breaking
2. [API] New check_token_balance RPC (Jan 8)
- User token balance endpoint
- Non-breaking
IdeaMachina Refactoring
For migrating IdeaMachina's evolution store from Zustand (client-side JSON blob) to Supabase relational tables, see the comprehensive guide:
references/idea-machina-evolution.md
Quick summary:
- Current state:
projectEvolutionStore(Zustand) persists evolution data as a JSON blob tostore_synctable. 30 files depend on it. 10 store versions (9 migrations). - Target state: Relational tables in
ai_promptschema (pm_evolutions,pm_sparks,pm_cores,pm_directions,pm_customer_personas,pm_force_modules) with RLS, queried via React Query. - Migration strategy: 3 phases — (1) schema + dual-write, (2) switch reads to DB, (3) remove Zustand store.
Use Pattern 7 in references/patterns.md for the general Zustand→DB migration pattern.
Related Skills
| Skill | Use For |
|-------|---------|
| code-wizard | Find code before refactoring |
| docs-updater | Sync docs after changelog |
| supabase-migration-writer | Database migrations |
| admin-panel-builder | Admin page refactoring |
| idea-machina | IdeaMachina feature context and evolution pipeline |
References
- Refactoring patterns: See references/patterns.md
- Changelog examples: See references/changelog-examples.md
- IdeaMachina evolution migration: See references/idea-machina-evolution.md