AgentSkillsCN

nobody-receives-code-review

适用于在收到代码审查反馈后、于落实建议之前使用——需秉持技术严谨与事实核查,而非流于表面的“同意”。

SKILL.md
--- frontmatter
name: nobody-receives-code-review
description: Use when receiving code review feedback, before implementing suggestions — requires technical rigor and verification, not performative agreement

Receiving Code Review

Overview

Code review feedback requires critical evaluation, not blind acceptance. Verify each suggestion technically before implementing.

The Rule

For each piece of feedback:

  1. Understand — What specifically is the concern?
  2. Verify — Is the concern technically valid? Check the code.
  3. Evaluate — Does the suggested fix actually address the concern?
  4. Respond — Agree with evidence, disagree with evidence, or ask for clarification

Response Types

Agree (with evidence)

code
You're right — line 42 doesn't handle the null case.
Fix: added null check with test in abc123.

Disagree (with evidence)

code
This is actually handled by the validation in middleware.py:30
which runs before this function is called. Here's the test
that covers it: test_middleware_validates_input.

Clarify

code
I'm not sure I understand the concern about the retry logic.
Are you worried about the backoff strategy or the max retries?

Anti-Patterns

BadGood
"Good point, fixed!" (without checking)Verify the concern, then fix
Implementing all suggestions blindlyEvaluate each on merit
Dismissing without explanationProvide evidence for disagreement
"I'll fix it later"Fix now or explain why not
Changing unrelated code during reviewOnly address review feedback

Red Flags

  • Agreeing with everything without verification
  • Making changes you don't understand
  • Getting defensive instead of curious
  • Implementing suggestions that introduce new bugs
  • Not running tests after applying feedback