Skills
97,501 packaged capabilities and playbooks for agents. Install any of them with one command.
Agent workflows
Turn a task or prompt into a deeper version of itself, not a bigger one. Before rewriting, say what a top professional in the field would do that the user isn't asking for, the rawest source material they would start from, the specific methods and tools they would use, and what (other than the user or the model) would show the result is right. Then rewrite the prompt to ask for exactly that. USE WHEN: /be-ambitious, "be ambitious", "make this prompt deeper", "what would an expert do here", "raise the bar on this task", "rewrite my prompt like a pro would ask it". NOT FOR: expanding scope, adding features, or general prompt polishing for tone or length
CCPM - spec-driven project management: PRD → Epic → GitHub Issues → parallel agents → shipped code. Use this skill for anything in the software delivery lifecycle: writing a PRD (“write a PRD for X”, “let's plan X”, “scope this out”), parsing a PRD into an epic, decomposing an epic into tasks, syncing to GitHub (“sync the X epic”, “push tasks to github”), starting work on an issue (“start working on issue N”, “let's work on issue N”), analyzing parallel work streams, running standups (“standup”, “run the standup”), checking status (“what's next”, “what's blocked”, “what are we working on”), closing issues, or merging an epic. Use ccpm any time the user is talking about shipping a feature, managing work, or tracking progress — even if they don't say 'ccpm' or 'PRD'. Do NOT use for: debugging code, writing tests, reviewing PRs, or raw GitHub issue/PR operations with no delivery context.
Make any software agent-native with HKUDS CLI-Anything — four routed modes: install ready-made harnesses via CLI-Hub (cli-hub list/search/info/install/launch), give agents the cli-hub-meta-skill for autonomous discovery, generate a new harness from any codebase/repo via the 7-phase /cli-anything pipeline, or iterate with :refine/:test/:validate; 40+ harnesses, 2,461 tests, Click CLIs with REPL + --json.
Design and build agent-first CLIs with HATEOAS JSON responses, context-protecting output, and self-documenting command trees. Use when creating new CLI tools, adding commands to existing CLIs such as joelclaw, or reviewing CLI design for agent-friendliness. Triggers on 'build a CLI', 'add a command', 'CLI design', 'agent-friendly output', or any task involving command-line tool creation.
Use Orca's computer-use CLI to inspect and operate local desktop app windows through accessibility trees, screenshots, and safe UI actions. Use for desktop app interaction: list apps/windows, get app state, read visible UI, click controls, type, press keys, scroll, drag, set values, or perform accessibility actions. Also use for browser windows, webviews, Orca app UI, or other desktop UI. Triggers include "computer use", "orca computer", "read Spotify", "read Slack", "control/click/read in a desktop app", and "get app state".
Core standards for all GitHub workflow agents. Covers authentication, smart defaults, repository discovery, dual MD+HTML output, screen-reader-compliant HTML accessibility standards, safety rules, progress announcements, parallel execution, and output quality. Apply when building any GitHub workflow agent - issues, PRs, briefings, analytics, community reports, team management.
Comprehensive guide to installing, configuring, operating, and troubleshooting OpenClaw — a self-hosted, multi-channel AI agent gateway. Use when the user asks about OpenClaw setup, configuration, channel management (WhatsApp/Telegram/Discord/Slack/iMessage/etc.), model provider setup, Gateway operations, multi-agent routing, security hardening, troubleshooting, or any maintenance task related to their local OpenClaw installation. Also use when encountering errors from `openclaw` CLI commands or the Gateway daemon.
Complete OpenCode starter pack — an interactive installation portal for environment setup, model connections, MCP integration, and skill installation. Load when the user says “install the starter pack,” “opencode starter pack,” or “help me install the starter pack.”
Build and deploy autonomous AI agents using the OpenServ SDK (@openserv-labs/sdk). IMPORTANT - Always read the companion skill openserv-client alongside this skill, as both packages are required to build and run agents. openserv-client covers the full Platform API for multi-agent workflows and ERC-8004 on-chain identity. Read reference.md for the full API reference.
Use the public `orca` CLI to operate Orca-managed worktrees/workspaces, terminals, repos, automations, worktree comments, and the browser embedded inside the Orca app. Use when the user says "$orca-cli", "use orca cli", "Orca worktree/workspace", "child workspace", "spawn codex/claude in a workspace", "read/wait/send Orca terminal", "terminal send", "Orca browser", or "control the browser inside Orca". Prefer this over raw `git worktree`, ad hoc PTYs, Playwright, or Computer Use when the task touches Orca-managed state. Use Computer Use for browser windows, webviews, or desktop UI outside Orca's embedded browser.
Authors and structures professional-grade agent skills following the agentskills.io spec. Use when creating new skill directories, drafting procedural instructions, or optimizing metadata for discoverability. Don't use for general documentation, non-agentic library code, or README files.
Diagnoses and refactors existing SKILL.md files whose references or assets are ignored by agents. Applies Required Reading Routers, hard STOP directives, gist-tripwire trimming, and flat one-level reference structures so bundled context is loaded deliberately. Use when auditing or rewriting long skills that feel shallow despite deep reference files. Do not use for authoring new skills from scratch, non-skill Markdown, or documentation without SKILL.md front matter.
Analyze code review feedback or issues found during development, extract general rules, and update CLAUDE.md and detailed rule documents. Use when the user says “update rules,” “add coding rules,” “update claude md,” or when a coding pattern needs to be documented.
Implement QRSPI-planned OpenSpec changes with intelligent parallelization. Use when the user wants to apply a QRSPI change, implement tasks with parallelization, or says "apply this QRSPI change", "implement with parallelization", "run the parallel slices". This skill is specifically designed for changes created via accelint-qrspi that include "Parallelization Strategy" sections in tasks.md. It orchestrates parallel sub-agent execution for independent task slices using OpenSpec CLI workflows. Make sure to use this skill when the user mentions applying QRSPI changes, running parallel implementation, or working on changes with vertical slices.
Alibaba Cloud Tablestore Agent Storage Skill. Use for building and managing Tablestore-based knowledge bases with the `tablestore-agent-storage` Python SDK. Triggers: "知识库", "tablestore", "ots", "表格存储", "agent storage", "knowledge base", "向量检索", "文档上传", "文档导入", "知识库同步", "tablestore-agent-storage", "AgentStorageClient"
Build AI agents with AWS Bedrock AgentCore. Use when developing agents on AWS infrastructure, creating tool-use patterns, implementing agent orchestration, or integrating with Bedrock models. Triggers on keywords like AgentCore, Bedrock Agent, AWS agent, Lambda tools.
Check that the CAD app can start after the CAD plugin is installed or updated, in Codex, Claude Code, Claude Desktop or another agent app—confirm uv is installed, point the user to uv's official installation guide when it is not, and tell them how CAD appears in their app and when to restart it. Use right after the CAD plugin is installed or updated, or when CAD's viewer did not start.
Use when an agent needs to operate the user's real Chrome session: listing tabs, snapshotting the page into actionable @refs, clicking, filling, typing into rich editors, pressing keys, evaluating JS, capturing screenshots, and reading console/network buffers. All actions go through CDP and run on backgrounded tabs without stealing focus.
Get an INDEPENDENT second-opinion review from a separate Claude agent in a read-only worktree — findings only, never edits to your working copy. Reads the real diff (branch, a commit, or uncommitted) to build a focused, risk-targeted brief, launches the reviewer under worktree isolation with report-only instructions, then summarizes findings by priority with `file:line`, carrying an exclusion list forward on later rounds. Treats returned findings as candidates to verify locally, not truth; states an explicit clean result rather than inventing issues; keeps secrets out of the prompt and never lets the reviewer mutate code. Use only when the user explicitly asks for a Claude review.
Patterns and anti-patterns for using OpenAI Codex Goals — the persistent objectives feature introduced in Codex 0.128.0. Use this skill whenever writing, reviewing, or debugging a `/goal` invocation, deciding whether a task should be a Goal at all, drafting a research Goal that needs an evidence ledger, or diagnosing a Goal that completed against the wrong surface. Triggers on `/goal`, "Codex Goal", "Codex goals", "persistent objective", "evidence-based completion", "iteration policy", "blocked stop condition", or any user message describing a multi-turn Codex task with a defined finish line. Trigger even if the user doesn't explicitly mention Goals — if they're typing "/goal" or asking Codex to "keep going until X", this skill applies.
Automatically create multiple background tasks in Codex App for complex, parallelizable knowledge work or programming tasks, with configurable models and reasoning effort. The main Agent handles planning, delegation, integration, and acceptance. Use for multi-source research, multi-chapter content, complex Skills/PPTs, cross-module development, independent verification, or 2+ independent workflows. Also use when the user explicitly requests model routing, background Workers, an Agents Team, or persistent project collaboration. Do not trigger automatically for simple Q&A, status checks, small single-file edits, strictly sequential tasks, or publishing/payment/deletion/account operations.
Lightweight shared entry point for the DingTalk MultiSkill. Use when the user refers generically to DWS/DingTalk operations without naming a product, requests cross-product orchestration, needs URL type pre-checks, or requires disambiguation of product boundaries. Clear single-product operations should prefer the corresponding dingtalk-* sub-skill; this skill only provides the global execution contract and on-demand reference navigation, not the full set of product commands.
Explain and manage a host-mode duoduo installation after onboarding. Use when the user asks how duoduo works, how stdio/daemon/channel/session fit together, where duoduo stores config and state, how to inspect current setup, how to upgrade duoduo itself, where something lives on disk (kernel_dir, runtime_dir, .env, descriptor.md), how to archive or recover a specific session (`duoduo session archive`, sessions-archive directory, restoring an archived session), or for broad 'configure duoduo' / 'fix my duoduo' requests that haven't narrowed to a specific channel or runtime setting yet. Also trigger for Chinese: 帮我理解 duoduo, duoduo 是怎么工作的, 看看我现在的 duoduo 配置, 帮我管理 duoduo, 升级 duoduo, 升级要注意什么, duoduo 哪个路径存什么, 归档 session, 删掉 session, 恢复归档 session.
A step-by-step problem-solving loop (classify the ask, define done, gather evidence, decide, act surgically, verify by observation, report outcome-first). Use when the user says "/fable-method", "use the fable method", or "approach this like Fable", or proactively when starting any multi-step task that no task-specific skill covers. Subcommands - plan (stop after the plan), audit (grade finished work against the loop), report (rewrite an answer outcome-first).
Discover, install, and manage AI agents across 43+ coding CLIs and IDEs from one source. They are automatically translated into each client's native format, not hand-ported, using the `agentshq` CLI: `find`, `add`, `list`, `remove`, `check`/`update`, and `init` scaffolding, scoped by `-g` (global) and `--ide` targets. Explicit-user-only and precise: it never installs, removes, or updates agents the user didn't ask for. Use to find agents or manage an existing agent catalog.
Pitfalls, examples, and edge cases for Hermes Kanban workers. The lifecycle itself is auto-injected into every worker's system prompt as KANBAN_GUIDANCE (from agent/prompt_builder.py); this skill is what you load when you want deeper detail on specific scenarios.
Get an INDEPENDENT second opinion from the Kiro CLI run strictly READ-ONLY — a separate reviewer runtime over the real diff, with no write access to the repo. Reads the real diff (branch, a commit, or uncommitted) to build a focused prompt, optionally injects the stack's convention skills (`--skill`, since kiro can't load them itself), then launches kiro via the bundled helper with trust scoped to `fs_read,execute_bash` so it can grep/git to verify findings but never edit. Never uses `--trust-all-tools` (the auto-mode classifier dead-ends read-only tasks on it) or hand-rolls the prompt through the shell; keeps secrets out, treats findings as candidates to verify, and reports an explicit clean result. Use only on explicit user intent or when a user-invoked workflow composes it as the kiro reviewer.
A tutorial router that loads the relevant setup guide from keywords and provides step-by-step installation instructions. Covers setup and maintenance tutorials for claude-mem, CLAUDE.md, the Chrome Gemini sidebar, Claude Code + 智谱 GLM, and the ECC plugin. Trigger when users ask for a tutorial, how to configure or install something, or mention specific tool names such as claude-mem, ecc, glm, gemini sidebar, or CLAUDE.md along with a setup need. This skill only handles configuration tutorials for the five local tools listed above. For web searches for tutorials, user experiences, or best practices, defer to laohan-sousuo (decided by Jeffrey on 2026-10-02; sousuo takes priority in conflicts).
Review the current session for durable learnings, then route each one through the five-way matrix: wiki page, guidance file, checkpoint, task annotation/plan, or discard. Use when the session uncovered decisions, architecture facts, commands, conventions, gotchas, or open questions that future sessions should inherit. Not for source ingestion or correcting stale wiki claims; use /loam::adding-to-memory or /loam::amending-memory. Routes goal-specific progress to /loam::setting-goals.
Shared loop-engineering reference for COG skills - the agent loop, deterministic verifiers, termination conditions, in-loop context management, and named patterns. Invoke when designing or debugging a skill that iterates (search-verify-retry, scan-until-dry, fetch-retry-gate).
Launch and manage the OpenClaw Dashboard web UI for monitoring all your bots, agents, models, and sessions. Automatically detects OS, checks for updates, installs dependencies, and starts the development server. Supports macOS, Windows, and Linux with automatic git detection and fallback download methods. Triggers include "打开机器人大盘", "打开 OpenClaw-bot-review", "打开Openclaw大盘", "打开 bot review", "open openclaw dashboard", "open OpenClaw-bot-review", "open bot dashboard", "open bot review", "launch bot review".
Update an OpenSpec change by revising its existing planning artifacts and keeping them coherent with one another. Use when the user wants to revise a change's plan, fold new decisions into it, or reconcile its artifacts after an edit. Never edits code.
The sole global workflow control plane for Phaser 4 game repositories; classifies and handles local work according to the current task scope and deterministic risk, and requires precise explicit approval only for side-effecting A4–A6 operations and their impact.
Make the agent show evidence when handing work over instead of just saying “done”. Use in three cases: when an agent is about to hand off work; when the user corrects the agent a second time about the same thing (“wrong again”, “how many times have I said this”, “how do I stop it from doing this again”); or when writing tests or cleaning up useless tests. prove it, stop repeating this mistake, clean up useless tests.
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.