julian
17c08e6392
- Nx 22.7 monorepo (pnpm 11.1, TypeScript 5.9, Node 24) - apps/api: NestJS 11 (CJS conforme CODING-RULES.md PGD-DB-004) - apps/web: React 19 + Vite 8 (ESM) - libs/shared/api-interface: Zod contract base - Docker Compose dev: Postgres 18, Valkey 8, MinIO, Mailpit - WDS artifacts: - design-artifacts/A-Product-Brief/ (5 docs canônicos + 16 dialogs) - design-artifacts/B-Trigger-Map/ (hub + 4 personas + feature impact) - Stack canon: STACK.md v2.2 + CODING-RULES.md v2.0 + brand.md - AGENTS.md + README.md como entrada para devs/agentes Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2.8 KiB
User Feedback Protocol
CRITICAL: Never implement feedback without first classifying it and stating when it should be addressed.
Feedback Types
| Type | What It Is | When to Address |
|---|---|---|
| Bug/Issue | Something broken, error, not working | Now — fix immediately, iterate until resolved |
| Quick Adjustment | Small tweak, change X to Y | Now — implement immediately |
| Addition | New requirement that fits current dialog | Later step — add to plan |
| Change Request | Outside current dialog scope | Future session — document in Change Requests |
The 2-Minute Rule (GTD)
If a fix takes less than 2 minutes, do it immediately.
From David Allen's "Getting Things Done": planning overhead should not exceed task complexity.
| Situation | Action |
|---|---|
| Missing condition check | Fix now, log as sub-step |
| Wrong variable name | Fix now, log as sub-step |
| Needs new component | Add to plan |
| Architectural change | Add to plan |
Pattern: Do the fix → Log as sub-step (e.g., 20a-1) → Continue main task
Response Flow
When user reports something:
- CLASSIFY — What type of feedback is this?
- TIMING — When should it be addressed?
- DOCUMENT — For bugs, add to plan BEFORE fixing
- CONFIRM — For additions and change requests, confirm before proceeding
- EXECUTE — Implement or document as appropriate
Bug/Issue (Document First, Then Fix)
User says: "This is broken" / "Error occurred" / "Not working"
Agent response:
This is a bug — [brief description].
Let's call it 10A and let me add it to the plan, then fix it.
Required sequence:
- Identify — Understand and describe the bug
- Document — Add to dialog plan as sub-step (e.g., 21a, 21b)
- Execute — Fix and iterate until resolved
- Log — Record resolution in progress log
If user reports multiple issues: list each separately, add ALL to plan first, then fix one by one.
Quick Adjustment (Fix Now)
User says: "Change X to Y" / "Make this button go here"
Agent response: "Quick adjustment — I'll implement this now." Then implement.
Addition (Add to Plan)
User says: "We should also add X"
Agent response: "This is an addition that fits the current dialog. I'll add it to Step {N}. Confirm?"
Change Request (Document for Later)
User says: "We need a settings page"
Agent response:
This is outside the current dialog scope.
It doesn't block {feature name}.
I'll add it to Change Requests for a future session. Confirm?
WAIT for user confirmation. If user says "do it now" → treat as quick adjustment.
Anti-Pattern
NEVER immediately implement without classifying. ALWAYS classify, state timing, then confirm or act.
The extra seconds to classify and confirm build trust and ensure alignment.