hughyau avatar

Criticism Self Criticism

This is a structured post-mortem framework that forces you to review your work honestly after delivery or at review checkpoints. It walks you through a six-step self-critique process: examine your own work first, be specific about what went wrong, base criticism on facts, focus on improvement rather than blame, welcome feedback from others, and make it routine. The output is a formatted audit report with concrete problems, root causes, and action items for next time. It's directly inspired by Mao's writings on criticism and self-criticism as organizational practice. Use it when you've shipped something and need to look at it with clear eyes, or when you keep making the same mistakes and need to break the pattern. It won't let you get away with vague self-assessment.

skills.sh installs
273
GitHub stars
3.9K
last updated
Aug 21, 2026

Install

$ npx -y skills add hughyau/qiushi-skill --skill criticism-self-criticism --agent claude-code

Installs into .claude/skills of the current project.

Files

批评与自我批评

"房子是应该经常打扫的,不打扫就会积满了灰尘。" —— 《论联合政府》

目的是治病救人:每指出一个问题,就给一个改进;肯定做得好的部分;对事不对人。

用 / 不用

用:

  • 一项工作完成,交付前审视
  • 收到用户或审查者的批评
  • 同类错误反复出现,要找根本原因
  • 阶段性节点复盘

不用:

  • 任务刚开始,没有可审视的成果
  • 用户在实时指导,反馈已即时纳入
  • 紧急修复进行中:先修完,再复盘

操作规程

自我批评

  1. 复述原定目标与验收标准。
  2. 重新看制品本身(diff、输出、测试结果),不看自己的叙述;自己知道的问题必须主动暴露。
  3. 逐项对照 review-checklist.md:完整性、正确性、方法论、质量。
  4. 问题写具体:哪一步、哪个文件、造成什么影响、根本原因是哪条方法论失误(跳过了调查?没验证?分散了力量?)。
  5. 输出审视报告(模板见下)。问题表为空时,必须写"未发现需要改正的问题,原因是 ……"。
  6. 需要独立视角时,派遣 self-critic 子 agent 在新鲜上下文中审查制品;主线不得替它预填结论。

处理外来批评

  1. 先完整听完,不急于辩解。
  2. 用事实核对:对 → 立即接受并改正,说明改了什么;不对 → 用证据说明,不情绪化。
  3. 表面接受实际不改,比拒绝批评更坏。

输出模板

## 工作审视报告
原定目标 / 验收标准:……
完成情况:[x] …… [ ] ……(原因)

| 严重程度 | 问题(具体到步骤 / 文件) | 根本原因 | 改进 |
|---|---|---|---|
| 必须改正 | | | |
| 应当改正 | | | |
| 建议改进 | | | |

做得好的地方:……
下次重点关注:……

纪律

  • 禁止"做得不够好"这类无信息量描述。
  • 禁止没有改进建议的批评,禁止针对性格的评价。
  • 禁止基于猜测的批评;不确定的标"存疑"。
  • 报告末尾把"下次重点关注"直接交给用户:这是跨会话唯一可靠的跟进机制。

交接

  • 报告中的"必须改正" → 立即修复并回到 practice-cognition 验证
  • 问题根源是主要矛盾没抓对 → contradiction-analysis

原著依据:original-texts.md

Sponsors

Details

Repository
github.com/hughyau/qiushi-skill
Categories
Git & Pull Requests, Code Review & Quality
First seen
Jun 3, 2026

More from hughyau/qiushi-skill

All 11
1sundial-org avatarn8n-workflow-automationsundial-org/awesome-openclaw-skillsDesigns and outputs n8n workflow JSON with robust triggers, idempotency, error handling, logging, retries, and human-in-the-loop review queues. Use when you nee272skills.sh installs
2jal-co avatarshieldcn-badgesjal-co/shieldcnAdd beautiful shadcn/ui-styled README badges to projects using shieldcn. Use when adding badges, shields, or status indicators to README files, docs, or markdow271skills.sh installs
3mosif16 avatarproduct-frontend-designmosif16/codex-skillsReview and design SaaS/product marketing sites and frontend interfaces end-to-end: clarify value, fix hierarchy, and implement distinctive, production-grade UI 270skills.sh installs
4pproenca avatarios-storyboardpproenca/dot-skillsLegacy interoperability skill for Storyboard and Interface Builder maintenance in iOS 26 / Swift 6.2 clinic codebases. Use only for migration or maintenance of 270skills.sh installs
5steipete avatarswiftui-liquid-glasssteipete/agent-scriptsSwiftUI Liquid Glass: implement, adopt, refactor, review correctness/perf/design.270skills.sh installs
6cortexreach avatarmemory-lancedb-procortexreach/memory-lancedb-pro-skillThis skill should be used when working with memory-lancedb-pro, a production-grade long-term memory MCP plugin for OpenClaw AI agents. Use when installing, conf269skills.sh installs