Treat review feedback as technical input to verify, not as a script to perform.
Context
receiving-code-review is the author-side companion to reviewer-side code-review. Its job is to stop two failure modes:
- blind implementation of feedback that breaks the real codebase
- performative agreement that sounds collaborative but skips technical verification
In Prodcraft, review reception must preserve brownfield constraints, existing contracts, release boundaries, and upstream decisions. External feedback is useful input, not automatic truth.
Inputs
- review-report -- The reviewer comments, blocking issues, questions, and requested changes.
- source-code -- The current implementation and surrounding codebase reality that the feedback must fit.
- test-suite -- The existing safety net used to verify each accepted change.
Process
Step 1: Read the Whole Review Before Acting
Do not start implementing the first comment immediately. Read the full review and group items into:
- blocking corrections
- clarifications needed
- technically questionable suggestions
- optional improvements
If multiple comments appear related, keep them together until the interaction is understood as a whole.
Step 2: Clarify Before Partial Implementation
If any review item is ambiguous, stop and ask before changing code. Do not implement the subset you understand while guessing at the rest.
Clarify when:
- the requested behavior is unclear
- the suggested scope is larger than the current slice
- the comment may conflict with an architecture or product boundary
- the reviewer appears to assume context that is not visible in the diff
Step 3: Verify Against Codebase Reality
For each review item, check:
- does the suggestion fit the current code and tests
- does it preserve brownfield compatibility and unsupported-flow rules
- does it violate YAGNI or introduce unused complexity
- does it conflict with reviewed architecture or contract decisions
- is the reviewer pointing at the real defect, or only at a local symptom
If the feedback is correct, implement it. If not, respond with technical reasoning and evidence instead of social agreement.
Step 4: Respond Factually, Not Performatively
Do not write praise or social filler such as:
- "You're absolutely right"
- "Great point"
- "Excellent feedback"
Instead:
- restate the technical issue
- say what changed
- say what remains unclear
- say why the suggestion should not be applied when evidence contradicts it
Step 5: Implement One Verified Item at a Time
Apply accepted feedback in a controlled order:
- blocking correctness or security fixes
- scope-safe cleanups
- optional or stylistic improvements last
Run the relevant tests after each meaningful change. If one item reopens another, record that explicitly instead of silently batch-editing everything at once.
Step 6: Produce a Review Response Record
Produce a review-response-record that captures:
- each review item
- whether it was implemented, challenged, or clarified
- the evidence used to decide
- the tests or checks run after accepted changes
- any remaining disputed items that still need reviewer or user resolution
Outputs
- review-response-record -- Item-by-item disposition of review feedback, with evidence, implementation status, and remaining questions.
Quality Gate
- Every review item is classified before implementation begins
- Ambiguous feedback is clarified before partial implementation
- Accepted suggestions are verified against current codebase reality
- Pushback, when needed, is technical and evidence-based
- Relevant tests are rerun after accepted changes
- The final response states what changed and what remains unresolved
Anti-Patterns
- Performative agreement -- sounding agreeable before verifying whether the comment is even correct.
- External-reviewer obedience -- treating outside feedback as an order instead of a hypothesis.
- Partial understanding, partial implementation -- fixing the easy comments while the important ambiguous ones remain unclear.
- Batch-and-pray -- applying many comments at once with no per-item verification.
- Context amnesia -- implementing a suggestion that breaks an upstream architecture, contract, or compatibility constraint.
Reference Material
For common feedback-handling traps, see Gotchas.
Related Skills
- code-review -- produces the reviewer-side findings that this skill processes
- tdd -- supplies the test discipline used to verify accepted review changes
- verification-before-completion -- verifies that claimed review follow-up is actually complete
Distribution
- Public install surface:
skills/.curated - Canonical authoring source:
skills/05-quality/receiving-code-review/SKILL.md - This package is exported for
npx skills add/updatecompatibility. - Packaging stability:
beta - Capability readiness:
beta