Loading...
Loading...
Found 13,522 Skills
Briefing process before starting a new task: ask only the essential questions (one at a time, look up what can be looked up first) → produce a first-version plan (no implementation) → submit for review → integrate and confirm with the user → also output an optimized task prompt as the foundation for this work → only start after confirmation → when execution ends, deliver a closing report that maps back to the plan. Triggered when the user says 'new mission', 'mission brief', 'brief me first', 'plan before doing', 'ask me first', 'kickoff', or the Chinese equivalents 「新任務」「任務簡報」「開工簡報」「先問我再做」「先給計畫」「這個任務怎麼開始」; also proactively propose running it when receiving a new task with more than three steps, or one where a mistake is hard to reverse. ⚠️ Does not apply to small things that take one or two steps, are reversible, and cost only a few minutes to redo — just do them directly. English triggers: 'new mission', 'mission brief', 'brief me first', 'plan before doing', 'ask me first', 'kickoff'.
Dylan's agent style on top of pstack: concise verified delivery, root causes over symptom patches, The Algorithm before design, plain /bro replies, live UI proof, reuse/simplify, and hard merge gates. Use for Dylan, /dyl-mode, or his style. "Get PR green" → /dyl-ready-pr. "Review this PR like me" → /dyl-review. Figma URL → /build-figma.
叶童本地健身会员业务的技能总入口。用户说“安装好了”“展示全部技能”“打开菜单”“怎么用”“更新叶童技能”，或提出小红书内容、每日热点选题、新课程朋友圈推广和业绩管理需求时使用；核验安装与三类能力状态，匹配已登记模块。支持包内更新；不处理教同行获客，未上线方向不冒充可用。
Route work that is too large for a single chat into a structured plan. Use when the user describes a multi-step build, a migration, a rewrite, an audit, a refactor across many files, or a research question with several threads. Use when a goal would otherwise need step-by-step supervision across more than one session. Also use when the user says "plan this", "where do I start", "help me build X", "this is a big one", or hands over a spec, a ticket, or a document and asks what to do with it.
Plan Japanese study, create printable learning materials, explain familiar expressions, and review exercises while maintaining a private learning history across sessions. Use for ongoing Japanese learning and study records, not unrelated document work.
Design, write, and validate chapter-based learning materials (<topic>-book/01-quickstart.md, 02-….md …) that cannot fit into a single quick-study guide, tailored to the reader persona. verify-book.mjs checks the mapping between per-chapter learning objectives and comprehension checks, the ordering that prevents concepts from being used before they are introduced, chapter reading-time budgets, exercises that “fail when incomplete and pass with the answer,” and the chapter dependency graph generated from book.json; explainer’s verify-doc.mjs checks each chapter’s outputs, diagrams, and HTML. Use when the user asks for a book, a tutorial series, a course, “learning materials,” “tutorial,” “hands-on,” “turn it into a book,” “organize it into chapters,” “start with 01-quickstart,” or when an explainer crash course would exceed ~20 minutes or needs exercises.
Print a friendly greeting from the Chalk plugin marketplace, used to verify plugin installation.
Prune low-value questions while preserving decisions that belong to the user. Use only when the user explicitly invokes $prune-the-tree.
Expand a project's decision space through informed questioning, uncovering relevant perspectives, explaining unfamiliar concepts, and tracing the implications of your answers.
Install a prebuilt DexSQL binary and connect it to the user's local MCP client. Use when asked to install, set up, or update DexSQL MCP; also accepts a user-provided local DexSQL executable.
Use when continuing, resuming, locating, reading, inspecting, auditing, or reviewing a previous AI coding-agent session, handoff transcript, chat log, exported conversation, saved artifact set, or session summary, on any platform (Claude Code, Codex, Cursor, Antigravity, OpenCode) or across several platforms in one ask, such as reviewing threads across Claude and Codex.
Independent translation-technique training for workplace learners. Two modes: Coach mode (Filtering → Deverbalization → Logic Surfacing → Re-expression → Cultural Adaptation, with the coach guiding each step) and Quick mode (the AI produces the translation directly, then explains step by step what adjustments were made and why). The Deverbalization stage adapts to the learner’s proficiency: the higher the proficiency, the less deverbalization is required. Trigger phrases: “translation technique training,” “translation training,” “translation practice,” “translation training.”