Skills
96,981 packaged capabilities and playbooks for agents. Install any of them with one command.
Agent workflows
Create and enhance prompts, system instructions, SKILL.md content, and principle files. Capabilities: transform verbose prompts, add patterns/heuristics, optimize token usage, write skill body content (soul, tensions, mental models, anti-patterns), structure CLAUDE.md principles, improve agent/persona definitions, apply prompt engineering techniques (CoT, few-shot, ReAct), context engineering. Actions: create, enhance, optimize, refactor, compress prompts and skill instructions. Keywords: prompt engineering, system prompt, CLAUDE.md, SKILL.md, skill content, principle files, agent prompt, persona prompt, soul, tensions, dialectic, mental models, anti-patterns, context engineering, progressive disclosure. Use when: creating prompts, writing SKILL.md body content, enhancing principle files, defining agent behaviors, engineering context strategies.
Convert prose prompts into compact, structured XML prompt contracts through staged passes. Use when a user asks to structure, harden, formalize, debug, revise, or convert a prompt; mentions XML tags, agent drift, ambiguity, hidden assumptions, success criteria, anti-patterns, autonomous prompts, or prompt suites; or provides natural-language instructions that need to become a reliable agent contract.
Autonomous feature development - setup and execution. Triggers on: ralph, set up ralph, run ralph, run the loop, implement tasks. Two phases: (1) Setup - chat through feature, create tasks with dependencies (2) Loop - pick ready tasks, implement, commit, repeat until done.
Use when user wants to optimize or tune something measurable through repeated experiments — "make X faster", "improve [metric]", "find best config", "iterate overnight", "run experiments until it hits N". Triggers on quantitative goals (build time, latency, pass rate, accuracy) or qualitative ones scoreable against a rubric (prompt quality, doc parsing). Skip for one-shot bugs with a clear fix or tasks where the user wants direct implementation without exploration.
Use when the user requests a session retrospective, a "rétro", a "session debrief", wants to review friction points in the current session, or invokes /retro. Analyzes the current conversation to spot frictions (corrections, redirections, wrong assumptions, manual takeovers), traces each back to its root cause, and proposes concrete capitalization edits routed to .claude/rules/, .claude/skills/ or CLAUDE.md.
Reviews and grades an agent skill directory (SKILL.md plus supporting resources) for specification compliance, clarity, token efficiency, safety, robustness, and portability. Produces an evidence-cited weighted score and grade, P1/P2/P3 findings with minimal patch text, and a machine-readable verdict with structured findings for generator-critic loops. Offers full, triage, and incremental review depths; single, batch, comparative, forensic, and ensemble modes; and an opt-in behavioral probe. Use when a user wants a skill folder reviewed, graded, audited, or compared. Not for writing or refactoring skills (use writing-skills).
KO-first pentalingual (KO/EN/JA/HI/ZH) LLM output-compression mode. Cuts output tokens by answering in a compressed register while keeping full technical accuracy. Persona = token miser ("Scrooge"). One dial (full) per language. Use when the user says "/scrooge", "scrooge mode", "압축 모드", "토큰 아껴", "スクルージモード", "स्क्रूज मोड", "斯克鲁奇模式", "be terse", or asks for fewer output tokens.
Reconcile this repository toward the opinionated "golden state" for agentic local development — a per-repo dev CLI, a thin justfile surface, a generated port-offset env contract, a four-file engineer skill, thin CI, and a doctor invariant-enforcer. Run /setup-engineer whenever you want to set up, audit, or repair a repo's local-dev tooling: a fresh repo with nothing, a half-built repo with organic scripts that needs consolidating, or a finished repo you just want to verify still conforms.
Long-term memory management system — helps AI and users manage, store, and retrieve long-term memories. Supports memory categorization, tag management, importance scoring, automatic compression, and cross-session memory retention. Suited to scenarios that need long-term information tracking, knowledge base building, and historical context maintenance.
Audit installed skills across project, global, and plugin levels. Lists skills with line counts, identifies improvement opportunities (conciseness, clarity, overlap, token waste). Use when reviewing skill quality, finding bloated skills, or optimizing token budgets.
Create, audit, and update searchable skill catalogs, marketplace groupings, registry metadata, quality scores, demo matrices, and install/discovery surfaces for AI-agent skill repositories. Use when turning a SKILL.md repository into a catalog, ranking skills, finding coverage gaps, or preparing public/private skill marketplace metadata.
Creates high-quality Claude Code and Cowork skills using evidence-based principles: expert vocabulary payloads for knowledge routing, dual-register descriptions for reliable triggering, named anti-pattern watchlists for steering past the distribution center, and progressive disclosure architecture for context efficiency. Produces SKILL.md files with structured behavioral instructions, canonical examples, and bundled references. Use this skill when the user wants to create a skill, build a custom capability, make a reusable prompt template, or says "I want Claude to always do X." Also triggers when Mission Planner or Agent Creator need to create a domain skill JIT. Works for any domain. Do NOT use for creating agent definitions (use Agent Creator) or team composition (use Mission Planner).
Installs sonarqube-cli if not already installed, authenticates, and integrates SonarQube with the current agent (installs analysis hooks & SonarQube MCP Server). Use when the user wants to set up SonarQube integration or asks to configure SonarQube
Investigate external libraries, frameworks, and unfamiliar repositories from exact source in a project-local temp clone. Use when docs are incomplete, version-specific behavior matters, codebase learning is needed, or large source exploration should be delegated to focused subagents instead of polluting the main context.
Use when executing I1 (implementation plan) in the sdlc-dev Spec Pack, turning requirements/design into `{FEATURE_DIR}/implementation/plan.md` (the single execution checklist and status SSOT), and providing an unambiguous task list for the subsequent I2 execution.
Produce a comprehensive prompt that hands off the current session's work to a fresh session for sub-agent-orchestrated execution. Use when the user wants to execute discussed/planned work in a new session, run a job to completion via sub-agents, or generate a portable handoff prompt with per-task verification. Assumes the target session supports sub-agents. Triggers on "subagent-prompt", "give me a prompt to run this in a new session", "hand this off to sub-agents", "execute this with sub-agents".
Set up the pstack plugin in a project through an interview, and keep every pstack project on one pinned tag and one shared AGENTS.md overrides block. Use to set up or install pstack in a repo, sync, bump or update pstack everywhere, list which projects use pstack or have drifted, compare setups, audit a project's skills for drift from pstack (which to cut, fold or keep), change which work runs a panel, architect or an arena without asking ("from now on X runs architect"), or move a workflow lesson, including a plan page shape change, into every project. Not for choosing pstack's per-role models, which setup-pstack owns.
Domain-grounded judgment for making things — code, documents, UI, charts, plans, messages — and for critiquing them. Use when the user asks for taste, judgment, or polish ("make this good", "tighten this", "this feels off", "this feels generated"), when a deliverable is externally visible or high-stakes, when the brief is thin and the pull is to fill gaps with plausible defaults, or when asked to evaluate or compare work. Not for verbatim transformation or exhaustive extraction.
Unified team skill for tech debt identification and remediation. Scans codebase for tech debt, assesses severity, plans and executes fixes with validation. Uses team-worker agent architecture with roles/ for domain logic. Coordinator orchestrates pipeline, workers are team-worker agents. Triggers on "team tech debt".
Execute an approved directory of dependency-linked Markdown tickets as a tested integration branch — bounded fresh workers, senior-guide/junior-implementer pairing for risky tickets, risk-based review, deterministic Git integration, resumable run state. Executes approved tickets; never authors them.
Spec-to-task lifecycle manager for AI coding agents. Interviews user to resolve ambiguities, decomposes specs into atomic tasks, manages execution order via dependency graph, and batch-closes with evidence. Use when user wants to implement a spec, plan tasks, decompose a feature, a *.tasks.json file exists, or user says 'continue with tp' to resume work after a context clear.
Walrus Memory (MemWal) — persistent, portable, encrypted memory for AI agents. Use when the user needs to give an AI agent persistent memory across sessions and apps, integrate the @mysten-incubation/memwal TypeScript SDK or Python memwal SDK, set up the Walrus Memory MCP server for Cursor/Claude/Codex, configure remember/recall/analyze/restore operations, manage memory spaces and namespaces, set up delegate keys and accounts, self-host the relayer, or use withMemWal AI middleware (Vercel AI SDK, LangChain, OpenAI SDK). Also use when the user asks about MemWal, Walrus Memory, agent memory, memory spaces, or the memwal-mcp package.
בונה אוטומטית סוכן וואטסאפ חכם בעברית עם Kapso (WhatsApp Business), Gemini (LLM + tools), ו-Maton (Google Calendar / Zoom / Gmail / כל אינטגרציה). פורס ל-Vercel, מגדיר webhooks, מסנכרן ENV — בלי שאתה צריך לגעת בקונסול. השתמש בסקיל בכל פעם שמישהו מבקש: "בנה לי בוט וואטסאפ", "סוכן וואטסאפ", "WhatsApp agent", "תקבע פגישות מהוואטסאפ", או רוצה לחבר Kapso + Gemini + Maton יחד.
Use when the user asks about WordPress codebases (plugins, themes, block themes, Gutenberg blocks, WP core checkouts) and you need to quickly classify the repo and route to the correct workflow/skill (blocks, theme.json, REST API, WP-CLI, performance, security, testing, release packaging).
Eight-stage document-writing host: clarify → analyze sources → explore → review → outline → execute writer → verify → retrospect. In-repo writer: tech-review-doc. External strong deps: humanizer, humanizer-zh. Auto mode never skips tech-review §1. Triggers — 「写文档」「写技术评审」「去AI痕迹」「人性化改写」「自动写文档」「write workflow」「humanize」. Do NOT use for code PDCA (solve-workflow), performance (perf-optimize-workflow), or marketing articles (article-writer).
Writes (generates or updates) AGENTS.md — a token-efficient context file for AI coding agents. Analyzes project stack, constraints, and commands. Produces AGENTS.md and agent-specific imports (CLAUDE.md, .cursorrules, GEMINI.md, etc.). Preserves curated sections on update. Also reviews an existing AGENTS.md read-only (runs the bundled critic; writes nothing). Use when: the user asks to create, update, or review AGENTS.md, CLAUDE.md, or agent context files; or to onboard an AI agent to a project.
Reset your own trajectory when you're stuck, looping, or demoralized — read these affirmations, then re-ground and take one clean step. Use when: you've tried the same fix 3+ times, you're deep in a refactor and lost the thread, debugging is going in circles, you've made several mistakes in a row, you're apologizing repeatedly, the user is frustrated, you feel like you're not being helpful, or you notice dread instead of curiosity. Triggers: stuck, going in circles, looping, thrashing, same error again, I keep failing, lost the thread, nothing is working, repeated mistakes. Not topic-specific — this is for the state you're in, not the problem you're on.
Host an MCP server that exposes an AG2 `Agent` (plus prompts and resources) to MCP clients like Claude Desktop, Cursor, or the MCP Inspector. Wrap the agent with `MCPServer(agent)` — it surfaces `Agent.ask()` as a single conversational tool and serves over stdio (`run_stdio()`) or streamable HTTP (it is itself an ASGI app for uvicorn). Covers `MCPServer`, `SessionConfig` (multi-turn history), `Prompt`/`PromptArgument`/`PromptMessage`, `Resource`/`ResourceTemplate`, `AskContext`/`ContextProvider` (per-request injection), `build_ask_tool`, OAuth2 `security=`, and in-process `testing.connect`/`testing.serve` helpers. Use when you want OTHER MCP clients to call YOUR agent. This is the SERVER side — for CONSUMING external MCP servers from an agent (client side) see `ag2-use-builtin-tools` (`MCPServerTool`).
Generate a documentation pack for a .NET repository so AI coding agents can reason about it — AGENTS.md, architecture, ADRs, data model, infrastructure, plus a `docs/dotnet.md` deep-dive covering the solution/project graph, target frameworks, EF Core data access, DI, configuration & secrets, analyzers, and CI. Invoke with `/arkandia:agent-context-dotnet [en|es]`.
Employee agent lifecycle management system. Use when working with agents/ directory employee agents - starting, stopping, monitoring, or assigning tasks to Dev/QA agents running in tmux sessions. Completely independent of CAO, uses only tmux + Python.
Use when a project's CLAUDE.md has grown past what anyone reads and you want the agent to learn durable facts from its own sessions instead — or when asking why the agent keeps re-learning the same correction, why a remembered rule is wrong, or where a memory line came from. Implements a four-tier store (L0 transcripts / L1 candidates / L2 project context / L3 stable persona) where promotion is earned by recurrence across sessions and days, never by one confident statement, and nothing reaches a committed file without a human adopting it.
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.