AI Skills
32 reusable skills for AI agents
Agent Development
ACP Router Skill: Route Coding-Agent Requests From One Plain-Language Prompt OpenClaw's official ACP routing skill. Receives plain-language requests for Claude Code, Cursor, Copilot, OpenCode or Gemini CLI and routes them into ACP runtime sessions or acpx-driven flows with thread-style conversation. MCP Builder Skill: Build High-Quality MCP Servers in Python or TypeScript Anthropic's official MCP development guide skill. Covers the full four-phase build: research and planning, protocol study, framework selection, and construction — FastMCP for Python, MCP SDK for Node/TypeScript. mcporter Skill: Configure, Authenticate, Call & Inspect MCP Servers OpenClaw's official MCP client skill. Uses the mcporter CLI to list servers, inspect tool schemas, and call tools directly over HTTP or stdio — one command into any MCP ecosystem. Skill Creator Skill: Create, Improve & Evaluate Agent Skills Anthropic's official skill-development skill. Covers creating, improving and measuring Agent Skills: draft, test, evaluate qualitatively and quantitatively, rewrite — plus description optimization for better triggering. Spike Skill: Run Throwaway Prototypes to Validate Feasibility Before Committing OpenClaw's official Spike skill. Answers "is this idea feasible" with the smallest runnable prototype — a five-step loop: question, research, build, stress, verdict (VALIDATED/PARTIAL/INVALIDATED). TaskFlow Skill: Coordinate Multi-Step Detached Work as One Durable Job OpenClaw's official TaskFlow skill. Coordinates multi-step detached tasks as one durable job with owner context, step state, waits and child tasks — resumable after restarts. ACP 路由器 Skill：一句话调度 Claude Code / Cursor / Gemini CLI 等编码 Agent OpenClaw 官方 ACP 路由 Skill。统一接收「用 Claude Code / Cursor / Copilot / OpenCode / Gemini CLI」等编码 Agent 请求，自动路由到 ACP 运行时或 acpx 会话，管理线程式长对话。 MCP Server 开发 Skill：Python 或 TypeScript 构建高质量 MCP 服务 Anthropic 官方 MCP 开发指南 Skill。覆盖 MCP 服务器四阶段全流程：调研规划、协议研读、框架选型、构建打磨，Python 走 FastMCP、Node 走 MCP SDK。 mcporter Skill：配置、认证、调用、检查 MCP 服务器与工具 OpenClaw 官方 MCP 客户端 Skill。用 mcporter CLI 列服务器、看工具 schema、直接调用工具，支持 HTTP 与 stdio 两种传输，一行命令接入任意 MCP 生态。 Skill 开发 Skill：创建、改进、评估 Agent Skills Anthropic 官方 Skill 开发 Skill。覆盖创建/改进/评估 Agent Skills 全流程：起草→测试→质与量评估→重写迭代，还能优化 description 提升触发准确率。 Spike Skill：跑一次性原型验证可行性，再决定要不要真做 OpenClaw 官方 Spike Skill。用最小可运行原型回答「这想法行不行」，五步循环：提问→研究→构建→压测→判定（VALIDATED/PARTIAL/INVALIDATED）。 TaskFlow Skill：把多步任务编排成可恢复的持久化作业 OpenClaw 官方任务流 Skill。把多步后台任务编排成单一 TaskFlow 作业，带归属人上下文、步骤状态、等待与子任务，重启后仍可恢复。
Content Creation
PDF Skill for AI Agents: Read, Merge, Split, OCR & Fill Any PDF Anthropic's official PDF skill. Lets AI agents extract text and tables, merge and split files, rotate pages, add watermarks, fill forms, encrypt and OCR scanned PDFs. DOCX Skill for AI Agents: Read, Edit & Create Word Documents Anthropic's official Word document skill. Lets AI agents read paragraphs and tables, create new documents, and edit styles and structure so .docx files flow through agent workflows. XLSX Skill for AI Agents: Read & Edit Spreadsheets Anthropic's official Excel skill. Lets AI agents read cell ranges, modify formulas and data, and create new spreadsheets — handling complex workbooks as easily as opening Excel. PPTX Skill for AI Agents: Create & Edit Presentations Anthropic's official PowerPoint skill. Lets AI agents read slide contents, create new presentations, and edit layouts and elements — turning "make a deck" into a single natural-language command. Diagram Maker Skill: SVG, HTML & Excalidraw for Architecture, Flows & Whiteboards OpenClaw's official diagram skill. Turns concepts, architecture and flows into artifacts, not prose: clean SVG for concepts, architecture SVG for software topology, Excalidraw for editable whiteboards. Frontend Design Skill: Distinctive, Intentional UI That Never Looks Templated Anthropic's official frontend design skill. Approaches UI as a small-studio design lead: ground the design in the subject, build a token system (color/type/layout/signature), and self-critique in two passes against "is this just the template answer". PDF 处理 Skill：让 AI 读写、合并、拆分、OCR 任意 PDF Anthropic 官方 PDF 处理 Skill。AI 可读文提取、合并拆分、旋转加水印、填表单、加密解密、扫描件 OCR，覆盖 PDF 全生命周期操作。 Word 文档处理 Skill：让 AI 读写、编辑、生成 DOCX Anthropic 官方 Word 文档 Skill。AI 可读取段落表格、创建新文档、编辑样式结构，让 .docx 文件在 Agent 工作流中顺畅流转。 Excel 表格处理 Skill：让 AI 读写、编辑电子表格 Anthropic 官方 Excel Skill。AI 可读取单元格区域、修改公式数据、创建新表格，处理复杂电子表格如同打开 Excel 一样顺手。 PPT 演示处理 Skill：让 AI 创建、编辑演示文稿 Anthropic 官方 PPT Skill。AI 可读取幻灯片内容、创建新演示文稿、编辑版式与元素，把「做 PPT」变成一句自然语言指令。 图表生成 Skill：SVG / HTML / Excalidraw 架构图、流程图、白板 OpenClaw 官方图表 Skill。把概念、架构、流程画成图而不是文字：clean SVG 用于概念图，architecture SVG 用于软件拓扑，Excalidraw 用于可编辑白板。 前端设计 Skill：打造有辨识度、不像模板的界面 Anthropic 官方前端设计 Skill。以「小型工作室设计总监」视角做界面：先从主题出发定调性，再定 token（色/字/布局/签名元素），两轮过稿自查「这是不是模板答案」。
Debugging
GitHub Skill: Issues, PRs, CI Logs, Releases & API in One CLI OpenClaw's official GitHub skill. Uses the gh CLI for issues, PRs, CI/check logs, comments, reviews, releases and repos, plus gh api for raw queries — automating the whole dev-collaboration loop. Node.js Debugging Skill: node inspect, CDP, Heap & CPU Profiling OpenClaw's official Node.js debugging skill. Starts with node inspect, escalates to CDP for scripted breakpoints, heap snapshots or CPU profiles; supports attaching to live PIDs, TypeScript breakpoints and child processes. Python Debugging Skill: pdb, breakpoint() & Remote Debugpy Attach OpenClaw's official Python debugging skill. Picks the smallest debugger per scenario: breakpoint() for source edits, pdb for launch-time, debugpy for remote or headless attach — from local scripts to stubborn prod bugs. GitHub Skill：issue / PR / CI / release / API 全流程管理 OpenClaw 官方 GitHub Skill。用 gh CLI 一站式处理 issue、PR、CI 日志、评论、评审、release 与仓库，gh api 直查任意接口，开发协作全流程自动化。 Node.js 调试 Skill：node inspect / CDP / 堆与 CPU 性能剖析 OpenClaw 官方 Node.js 调试 Skill。默认 node inspect 起步，需要脚本化断点/堆快照/CPU profile 时升级 CDP；支持附加已有 PID、TypeScript 断点与子进程调试。 Python 调试 Skill：pdb / breakpoint / debugpy 远程调试一条龙 OpenClaw 官方 Python 调试 Skill。按场景选最小调试器：breakpoint() 改源码、pdb 无侵入启动、debugpy 远程/无头进程附加，覆盖本地到生产疑难 bug。
Ops & Security
Host Healthcheck Skill: Audit & Harden an OpenClaw Host OpenClaw's official host healthcheck skill. Assesses host risk with read-only checks (SSH/firewall/updates/backups/encryption/gateway exposure), then proposes reversible, staged hardening — confirming access paths before any change. 主机体检 Skill：审计并加固 OpenClaw 主机安全 OpenClaw 官方安全体检 Skill。先只读评估主机风险（SSH/防火墙/更新/备份/加密/网关暴露），再给出可逆的分步加固方案，改动前先确认访问路径。