菜单
Systematically find problems in a design before building it. Four lenses: unnecessary complexity, missing fundamentals, feasibility gaps, and scope mismatch. Use after creating an implementation plan or when a design feels wrong. Invoked via /critique.
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/robert-chiniquy/dotfiles --skill critique命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
正在显示 SKILL.md
基于 SOC 职业分类
| name | critique |
| disable-model-invocation | true |
| description | Systematically find problems in a design before building it. Four lenses: unnecessary complexity, missing fundamentals, feasibility gaps, and scope mismatch. Use after creating an implementation plan or when a design feels wrong. Invoked via /critique. |
Apply four lenses to the design. Each lens asks different questions.
For each problem found:
Number problems sequentially. The list IS the deliverable.
Sweep open issue trackers, TODOs, active branches, and source evidence to find narrowly scoped, independent, verifiable tasks suitable for remote agents or parallel delegation. Use when asked to find pebbles, small delegatable work, fire-and-forget candidates, backlog slices, ready items, or pool/task combinations.
Write a scorecard — a dark, skimmable one-screen readout of the current situation in the user's work, rendered as a terminal (TUI) artifact by the `scorecard` binary from a structured markdown file. Use ONLY when the user explicitly asks for a scorecard (or a "readiness card" / "go/no-go card" / "coverage snapshot card"). A plain "status", "status?", or other status/situation question gets a prose answer, never this skill. NOT a web page — output is the dark TUI. Preserves older scorecards by datestamp. Triggers on: scorecard, readiness card, go/no-go card, coverage snapshot card.
Present comparisons as narrow ASCII bar charts in the terminal. Use when user asks to compare items, metrics, counts, or coverage between things.
