Review completed work and its process against the original goals and acceptance criteria. Identify specific problems, root causes, and actionable fixes, while checking external criticism against evidence and correcting valid issues.
批评与自我批评：在工作完成、阶段验收、收到批评或同类错误反复出现时，对成果和过程做诚实、具体、基于事实的审视，输出可执行的改进项，并处理外来批评而不辩解。触发信号包括 review、复盘、审查、"帮我看看有没有问题"、"你确定吗"；任务刚开始或只是单步查询时不触发。
English: Criticism and self-criticism. After delivery, at a review checkpoint, on receiving criticism, or when the same mistake recurs, examine the result and the process honestly and concretely, produce actionable fixes, and accept valid criticism without defensiveness. Triggers include review, retrospective, audit, "check this for problems", "are you sure"; skip at the start of a task or for one-step lookups.
Review the instructions the agent will receive before you install.
"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：
自我批评
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
## 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：……
practice-cognition 验证contradiction-analysis原著依据：original-texts.md
No reviews yet. Tell others what you thought.
No login needed. Share how it worked for you, ask a question, or note where it did not run.
No comments yet. Be the first to write one.
Guide an agent to ground judgments in evidence, separate facts from inferences and unknowns, verify results, and diagnose failures. It also selects a suitable methodology skill when investigation, prioritization, validation, or another structured approach would help.
Organize competing tasks by impact, difficulty, and dependencies, then choose one main priority. State what completion looks like, keep unrelated requests queued, and finish and verify the chosen task before reassessing the remaining work.
Break down a complex problem into opposing needs, identify the central conflict and its dominant side, then choose a response. The analysis also highlights changing priorities and tensions that need monitoring as work progresses.
Investigate firsthand sources before making a judgment or recommendation. Separate verified facts, reasoned inferences, and unknowns, document their sources, and produce a conclusion that explains the current situation, constraints, and remaining uncertainty.
Combine feedback from users, code, logs, and runtime results into a reasoned plan. Separate agreement from conflicting evidence, explain the judgment, and validate it with the people or systems that supplied the original information.
Plan decisions where improving one goal can harm another. Identify competing priorities, set the current emphasis and minimum acceptable levels, check effects across the whole plan, and define signals that call for rebalancing.