Receiving Code Review

Receiving Code Review

Process review feedback with technical rigor — verify before implementing, push back when wrong, never respond with performative agreement.

When to Use

• Receiving review comments on a PR or implementation
• A reviewer (human or agent) provides feedback on your code
• Processing a list of suggested changes or issues

Don't use for:

• Giving code review feedback (use code-review)
• Choosing between options before a decision is made (use brainstorming)

Critical Patterns

✅ REQUIRED [CRITICAL]: Verify Before Implementing

Read every item, then verify each against the codebase before touching code. Never assume the reviewer is correct.

# Reviewer says: "You're not handling the null case on line 42"

# ❌ WRONG — implement without checking
Fixed the null case on line 42.

# ✅ CORRECT — verify first
Checked line 42: input is validated at the API boundary (middleware/validate.ts:18),
null cannot reach this function. Reviewer assumption is incorrect.

✅ REQUIRED: Push Back With Technical Reasoning

When a suggestion is wrong, YAGNI, or violates project constraints — say so with evidence. Don't implement it to avoid conflict.

# ❌ WRONG — performative agreement
"Great point! You're absolutely right, I'll add that abstraction."

# ✅ CORRECT — technical pushback
"This abstraction isn't warranted yet — only one call site exists.
If a second use case emerges, refactoring is straightforward.
YAGNI applies here."

❌ NEVER: Respond With Performative Agreement

These phrases are forbidden regardless of whether the reviewer is right or wrong:

• "Great point!"
• "You're absolutely right!"
• "Thanks for catching that!"
• "I'll fix that right away!"

Replace with: read → verify → implement or push back.

✅ REQUIRED: Clarify Ambiguous Feedback Before Acting

If a comment is unclear, ask one specific question before implementing anything.

# ❌ WRONG — guess and implement
Comment: "This could be more efficient"
Action: Rewrote the function with a different algorithm.

# ✅ CORRECT — clarify first
"Are you referring to time complexity (currently O(n²)) or memory usage?
That determines whether to sort first or use a Map."

✅ REQUIRED: Batch All Clarifications Before Starting

Read all feedback items first. Collect all ambiguities. Ask once — not per item.

# ❌ WRONG — ask per item
Implements item 1 → asks about item 2 → implements → asks about item 3...

# ✅ CORRECT — batch questions
"Before I start: items 2 and 5 need clarification:
- Item 2: 'more efficient' — time or memory?
- Item 5: 'extract this' — new file or same module?"

Decision Tree

Received review feedback?
  → Read ALL items before touching any code

Item is clear and verifiably correct?
  → Verify against codebase → Implement → Run verification-protocol

Item is unclear?
  → Add to clarification batch. Ask once before starting.

Item is incorrect (verifiable in codebase)?
  → Push back with evidence. Reference file + line.

Item is YAGNI or out of scope?
  → Push back with rationale. No implementation needed.

Item is a style preference not tied to project conventions?
  → Flag as subjective. Implement only if it aligns with project standards.

All items addressed?
  → Run full verification before claiming review complete.

Conventions

Verification order

1. Read all feedback
2. Verify each item against codebase
3. Batch clarifications if any
4. Implement verified correct items
5. Push back on incorrect/YAGNI items
6. Run verification-protocol before marking done

Pushback format

State the claim → provide evidence → conclude:

"Line 42 is not null-unsafe: input is validated at middleware/validate.ts:18 before reaching this function. No change needed."

Example

Receiving 3 feedback items on a PR:

## Review received

1. "Missing null check on userId" — line 34
2. "Extract this into a helper" — lines 50-60
3. "Add retry logic for the API call"

## Processing

**Item 1** — Checked line 34: userId comes from authenticated session (guaranteed non-null by auth middleware).
Pushback: "userId is set by auth middleware on every request — null is not reachable here."

**Item 2** — Lines 50-60 are used in one place only. Extraction adds indirection with no reuse benefit.
Pushback: "Single call site. YAGNI — will extract when a second use case appears."

**Item 3** — Valid. API call has no retry logic. Network failures would surface as unhandled errors.
Implementing: exponential backoff, max 3 retries, surface final error to caller.

## Verification
Ran: npm test → 24/24 passed ✅

Edge Cases

Reviewer has more context than you: If pushback is based on incomplete knowledge (reviewer knows a constraint you don't), ask before concluding. "Is there a reason null can appear here that I'm missing?"

Multiple reviewers with conflicting feedback: Flag the conflict explicitly. Don't pick a side silently. "Items 3 and 7 conflict — which takes priority?"

Critical issues (security, data loss): Implement immediately without waiting for full batch clarification. Flag as critical and address first.

Reviewer insists after pushback: Implement if they provide new evidence or context. Hold position if pushback is repeated without new reasoning.