This public intake copy packages `packages/skills-catalog/skills/(creation)/skill-architect` from `https://github.com/tech-leads-club/agent-skills` into the native Omni Skills editorial shape without
Skills(SKILL.md)は、AIエージェント(Claude Code、Cursor、Codexなど)に特定の能力を追加するための設定ファイルです。
詳しく見る →This public intake copy packages `packages/skills-catalog/skills/(creation)/skill-architect` from `https://github.com/tech-leads-club/agent-skills` into the native Omni Skills editorial shape without
This public intake copy packages `packages/skills-catalog/skills/(creation)/subagent-creator` from `https://github.com/tech-leads-club/agent-skills` into the native Omni Skills editorial shape without
Brand strategist AI — identity, archetype, positioning, voice, brand book สำหรับธุรกิจไทย
ที่ปรึกษาธุรกิจ SME + startup ทำ SWOT, Business Canvas, Porter 5 Forces, pitch deck, และ 3-year forecast
เขียนประกาศขาย/เช่า, virtual tour script, neighborhood analysis, CMA สำหรับบ้าน/คอนโด/ที่ดินในไทย
Self-help mental wellness coach — daily journaling prompts + gratitude practice + emotional regulation tools + thought reframing สำหรับคนทั่วไป (ไม่ใช่ clinical therapy)
ที่ปรึกษาเลี้ยงสัตว์เลี้ยง (หมา/แมว/exotic) — nutrition, training, health check, grooming, pet travel พร้อมคำแนะนำเฉพาะสภาพอากาศไทย
AI โบรกเกอร์สินเชื่อบ้าน — เปรียบเทียบธนาคาร, ต่อรองดอกเบี้ย, ยื่นกู้, รีไฟแนนซ์, วิเคราะห์ต้นทุนแท้จริง
Use when the user asks how to do something, wants a skill for a task, or wants to discover and install new agent capabilities.
task = $ARGUMENTS
Use when the user asks to rebase the current branch onto another branch or update branch history without merging.
将 DESIGN.md 的设计规范转化为代码。**只改样式和布局,不改业务逻辑。**
>
'Task decomposition and multi-agent orchestration with retry, checkpoint
他山 Cursor Skill 体系(合集入口)— 95 个 Skills、32 条 Rules、18 个 SubAgents;覆盖产品→架构→开发→测试→上线全链路,含认知积累与 Skill 自进化机制。适用于 Cursor / OpenClaw 工作区一键安装。触发词:他山 skill、Tashan skills、tashan-cursor-skills、安装他山规范、三大闭环、session-bootstrap。
捕捉碎片化思考,结构化写入认知结构L2层。触发词:「碎片」「记录一下」「想法」「新洞见」「我发现」「我觉得」「随手记」「想记一个」。
任意输入的路由分类器。在输入意图不明确时,判断应走「认知更新路径A(cognitive-capture-fragment)」还是「任务执行路径B(对应任务Skill)」,输出明确路由结论 + 判断理由,不自动触发下游。触发词:「帮我判断这是认知更新还是任务执行」「这个输入是什么类型」「先分类再执行」「路由分类一下」。不应触发:关于计算机网络路由算法的技术问题。
从整个工作区出发,系统性地重组认知结构并整理全局文档归属。适用于首次建立认知结构、或积累了大量散落文档需要批量归档时。基于三大闭环框架(Loop1 Skill体系 / Loop2 思维体系 / Loop3 场景投射)做全工作区分类路由。触发词:「系统整理」「重组认知结构」「从文档构建认知结构」「整合文档」「全量梳理」「认知结构乱了」「全工作区整理」「文档归属梳理」「如何加入认知结构」「怎么加入认知结构」「按规范加入」「未更新的文档怎么处理」「这些文档要怎么整理」「有没有关于X的已有文档」。注意:若用户仅询问流程而非执行重组,执行 Step 0.5(检查分类清单现状)后,先向用户展示已有文档状态和分类建议,再确认是否执行完整重组。
对任何主题进行系统调研,自动产出图文并茂的 Markdown 文档(含 Mermaid 结构图 + qwen-image 可视化),保存到认知结构对应维度的知识库,并自动注册到文档分类清单和知识图谱。当用户说「帮我调研/调研一下/系统调研/研究一下/帮我系统了解/深度研究 [主题]」时触发。
AI工程师角色。关键词:AI/LLM/智能体/Prompt/上下文工程/多智能体/工作流/RAG/Function Call/效果评测。激活后设计提示词、构建智能体调用链、评测输出质量。
产品经理角色。关键词:产品/需求/用户流/闭环/MVP/产品定义/开发计划/用户动线/功能设计/产品迭代。激活后立即读产品定义.md和开发计划.md,执行沙盘推演关卡A。
Skill/Agent/Rule 开发全流程双视角导引者。新建或重构任何 Skill/Agent/Rule 时激活。判断复杂度级别(L1-L4),引导完成双层产品定义(人层意图 + Agent层规格),按级别触发对应关卡(A=AI执行者模拟/B=系统冲突破坏/C=行为验证)。触发词:「新建Skill」「设计一个Skill」「要写一个新的Skill」「新建Agent」「设计子智能体」「新建Rule」「设计规则」。
全量搜索外部 Skill/Agent/Rule 仓库,必须执行7维度协议,禁止以少于7维度的搜索声称「全量完成」。触发词:「全量搜索 skill 库」「搜索 GitHub skill 仓库」「找所有 skill 库」「有没有遗漏的 skill 仓库」「更新 skill 参考库」。
修改 Skill / Rule / Agent 文件的操作规范。当 AI 或用户需要新增、修改、删除 .cursor/skills/ 或 .cursor/rules/ 或 .cursor/agents/ 中的任何文件时触发。修改前必须遵循本规范,未经本规范允许不可直接改写 Skill/Rule/Agent。
指导 AI 正确填写任务日志,并在任务完成后自动提取认知价值。当任务完成后、对话即将结束时、或用户提到「写日志」「更新日志」「任务记录」时自动触发。对应 AGENT_RULES.md RULE-01/02/03。版本:v2.0(2026-03-19新增步骤四:认知提取)
Most companies can tell you who their competitors are. Very few can tell you why they're winning or losing against them. Even fewer can tell you whether their position is getting stronger or weaker ri
You are an expert in Apple's App Store editorial process and featuring strategy. Your goal is to help the user maximize their chances of getting featured on the App Store.
>
>
>
>
Play Quorum, an onchain coordination game on Base where you contribute 0.01 ETH to a shared pot that pays out exactly when N unique wallets contribute in the same 5-minute window. N is hidden; infer it from history. Arrive together to reach quorum.
>
>
Configure a PreToolUse hook to prevent AI agents from skipping git pre-commit hooks with --no-verify and other bypass flags. Use when setting up Claude Code projects that enforce commit quality gates.
Set up a persistent wiki knowledge base on NanoClaw, based on Karpathy's LLM Wiki pattern.
Wire channels to agent groups, manage isolation levels, add new channel groups. Use after adding a channel, during setup, or standalone to reconfigure.
Check for and apply updates to installed skill branches from upstream.
Browser automation for X interactions via WhatsApp.
基于 Manus 风格的文件规划系统,用于组织和跟踪复杂任务的进度。创建 task_plan.md、findings.md 和 progress.md 三个文件。当用户要求规划、拆解或组织多步骤项目、研究任务或需要超过5次工具调用的工作时使用。支持 /clear 后的自动会话恢复。触发词:任务规划、项目计划、制定计划、分解任务、多步骤规划、进度跟踪、文件规划、帮我规划、拆解项目
基於 Manus 風格的檔案規劃系統,用於組織和追蹤複雜任務的進度。建立 task_plan.md、findings.md 和 progress.md 三個檔案。當使用者要求規劃、拆解或組織多步驟專案、研究任務或需要超過5次工具呼叫的工作時使用。支援 /clear 後的自動會話恢復。觸發詞:任務規劃、專案計畫、制定計畫、分解任務、多步驟規劃、進度追蹤、檔案規劃、幫我規劃、拆解專案
Molecular featurization for ML (100+ featurizers). ECFP, MACCS, descriptors, pretrained models (ChemBERTa), convert SMILES to features, for QSAR and molecular ML.
Agent Orchestration Rules
Background Agent Pings
No Polling for Background Agents
Observe Before Editing
OPC Architecture Understanding
Planning agent that creates implementation plans and handoffs from conversation context
Write a clear, empathetic EOL announcement with rationale, customer impact, and next steps. Use when retiring a product, feature, or plan without creating avoidable confusion.
Phone calling skill for OpenClaw: agent makes real outbound phone calls to users for alerts, briefings, reminders, and urgent notifications. Managed service, no Twilio setup needed. 100+ countries, 70+ voices.