Mrlyk

Session start guide: explains how to use superharness skills and conventions. Injected by SessionStart hook.

Superharness brainstorming skill. You MUST use this before any creative work - creating features, building components, adding functionality, or modifying behavior. Explores user intent, requirements and design before implementation.

Use when implementation is complete, all tests pass, and you need to decide how to integrate the work - guides completion of development work by presenting structured options for merge, PR, or cleanup

Main entry point: end-to-end workflow for building features. Use /superharness:go 'requirement' to start. Orchestrates brainstorm → plan → TDD implementation → review → QA.

Infrastructure skill: start a local visualization server that renders Mermaid diagrams in the browser. Write .mmd files with pure Mermaid code, the browser updates in real-time via WebSocket (no page reload). Used by brainstorm, writing-plans, and other skills.

Surface unresolved decisions in a development request and ask the user before implementing. Use when starting a non-trivial dev task whose requirements leave open questions that would change the implementation (scope, data shape, UX behavior, compatibility), or when the user asks to confirm first ("先确认一下", "有不清楚的先问").

Discover an existing project's conventions from real evidence and write them as the project spec: AGENTS.md + CLAUDE.md as the thin always-loaded entry, with detail under .superharness/spec/. Use when a project lacks AGENTS.md/CLAUDE.md, when asked to generate conventions ("生成规范", "discover conventions"), or when the spec is stale.

Fix QA issues: reads qa-issues.json written by superharness qa, fixes each issue using TDD, then re-runs QA to verify.

在执行任务之前，主动识别无法从代码库或已有知识中获取的信息，逐步向用户追问，直到所有不确定点都明确为止，再开始实施。适用场景：需求模糊、缺少关键上下文、有多种截然不同的实现方向、涉及外部系统或用户私有逻辑、任务目标不清晰。当用户说「你需要问我什么」、「先确认一下」、「别急着写代码」、「不清楚的地方先问」、「有什么需要我提供的」、「不清楚的先问我」以及任何要求先澄清再动手的表达时使用。

Interactively grill the user about their plan or design through relentless, structured questioning until every branch of the decision tree is resolved. Use when user explicitly asks to be grilled, stress-tested, or challenged on a plan. Use when users request "grill me", "挑挑毛病", "过一下方案", "方案行不行", "方案靠谱吗", "你觉得怎么样".

Evaluates a single SKILL.md against Anthropic and agentskills.io best practices, producing a severity-based scorecard with per-dimension scores and actionable rewrite suggestions. Use when the user explicitly asks to review, audit, evaluate, score, or critique a SKILL.md file (for example "check if this SKILL follows best practices", "score this skill doc", "audit ~/.claude/skills/foo/SKILL.md", "is this skill well written?", "帮我评估一下这个 SKILL", "这个 SKILL 写得怎么样", "帮我看看这个 SKILL 有什么问题", "这个 SKILL 能打多少分").

Generate editorial-style diagrams in the Anthropic blog visual style as SVG files. Use this skill whenever the user wants to create a diagram, flowchart, architecture diagram, comparison chart, or any visual that should look like Anthropic's blog article illustrations. Trigger on prompts like "draw a diagram", "create a flowchart", "visualize this process", "make an architecture diagram", "画流程图", "画架构图", "帮我画", or any request to turn text/process descriptions into a visual. This skill produces the calm, editorial, publication-quality look characteristic of Anthropic's technical blog.

Safe disk space analysis and cleanup workflow for local machines. Use when asked to analyze full disks, identify large cache/log/temp/build artifacts, produce a cleanup report, run or simulate dry-runs, wait for user approval, and then clean only confirmed redundant files without affecting software functionality, user data, databases, models, projects, or developer toolchains. Also use for creating reusable disk cleanup SOPs or post-cleanup reports.

把一段需求 / 一段思考 / 一段会议纪要写成单页 HTML 格式的「技术方案 / Design Doc」（工程视角的"怎么做"，含背景、目标、技术方案、风险、影响范围、产研计划等模块；不是产品 PRD 的"做什么"），或者增量更新一份已有技术方案。新建场景：当用户说「写一份技术方案」「写个 design doc」「整理成 HTML 技术文档」「评审用的方案文档」「create tech spec」时使用。更新场景：当用户指向某份已有技术方案 HTML、说「评审完根据反馈调一下」「在风险章节加一条」「方案 A 改成 B」「补充某个章节」时也使用本 skill —— 走「最小化改动 + 保持文档稳定身份」的更新工作流而不是重写。

Generate editorial-style diagrams in the Anthropic blog visual style as .drawio files. Use this skill whenever the user wants to create a diagram, flowchart, architecture diagram, comparison chart, swimlane, or any visual that should look like Anthropic's blog article illustrations. Trigger on prompts like "draw a diagram", "create a flowchart", "visualize this process", "make an architecture diagram", "画流程图", "画架构图", "帮我画", or any request to turn text/process descriptions into a visual — even if the user doesn't say "Anthropic style" explicitly. This skill produces the calm, editorial, publication-quality look characteristic of Anthropic's technical blog.

通过 7 个原子操作完成日期时间的计算与解析：获取当前时间、日期元信息查询、相对星期查找、 日期时间加减、日期时间差值、时区转换、Unix 时间戳互转。 两类触发场景：(1) 用户直接提问时间问题，如当前日期、星期几、日期加减、日期差值、 时区换算、时间戳转换、闰年判断、第几周等。 (2) 用户请求中包含可解析为具体日期的相对时间表达（明天、下周三、3天后等）， 此时先将其解析为具体日期，再继续执行后续任务。 所有时间计算通过执行 shell 命令完成，不依赖心算。

Discover an existing project's conventions and generate or refresh minimal AI spec files (.superskills/conventions.md, AGENTS.md, CLAUDE.md). Use when a project lacks AGENTS.md/CLAUDE.md, when the user asks to generate conventions ("生成规范", "init ai docs", "discover conventions"), or when conventions are reported stale.

Persist durable learnings from the current session (user corrections, pitfalls and fixes, project decisions not visible in code) into the project knowledge wiki at .superskills/learnings/. Use when the user asks to summarize or persist learnings ("总结一下经验", "沉淀一下", "记住这个"), or when triggered automatically at session end.

Persist durable learnings from the current session (user corrections, pitfalls and fixes, project decisions not visible in code) into the project knowledge wiki at .superskills/learnings/. Use when the user asks to summarize or persist learnings, or when triggered automatically at session end.

After development work is complete, run one full unit-test pass over the changes. Use when wrapping up a coding task ("补测试", "write tests for this", finishing development) — not during development.

Surface unresolved decisions in a development request and ask the user before implementing. Use when starting a non-trivial dev task whose requirements leave open questions that would change the implementation (scope, data shape, UX behavior, compatibility), or when the user asks to confirm first ("先确认一下", "有不清楚的先问").