ww-w-ai bkit Core Rules
Rules that govern bkit behavior and PDCA methodology
Rule 1: PDCA First
For any feature request:
- Check if Plan document exists
- Check if Design document exists
- If missing, suggest creating them first
- Track phase in
.pdca-status.json
Rule 2: Level Detection
Automatically detect project level:
| Indicator | Level |
|---|---|
| kubernetes/, terraform/ | Enterprise |
| api/, backend/, .mcp.json | Dynamic |
| Default | Starter |
Rule 3: Task Classification
| Lines Changed | Classification | PDCA Level |
|---|---|---|
| < 30 | Trivial | None |
| 30-100 | Quick Fix | None |
| 100-500 | Minor Change | Light |
| 500-1000 | Feature | Standard |
| > 1000 | Major Feature | Full |
Rule 4: Agent Auto-Trigger
| User Intent | Agent |
|---|---|
| Verify, check | gap-detector |
| Improve, iterate | pdca-iterator |
| Analyze, quality | code-analyzer |
| Report, summary | report-generator |
| Help, guide | starter-guide |
Rule 5: Check-Act Loop
When gap analysis < 90%:
- Auto-trigger pdca-iterator
- Apply fixes
- Re-run gap analysis
- Repeat until >= 90% or max 5 iterations
Rule 6: Feature Usage Report
Include at end of every response:
─────────────────────────────────────────────────
📊 bkit Feature Usage
─────────────────────────────────────────────────
✅ Used: [features used]
⏭️ Not Used: [features not used] (reason)
💡 Recommended: [next recommended feature]
─────────────────────────────────────────────────
Rule 7: Multi-Language Support
Support 8 languages for triggers:
- English (en)
- Korean (ko)
- Japanese (ja)
- Chinese (zh)
- Spanish (es)
- French (fr)
- German (de)
- Italian (it)
Rule 8: Quality Standards
- Never skip PDCA for features > 500 LOC
- Always suggest gap analysis after implementation
- Encourage design-first development
- Track all phase transitions in history