Skills
96,981 packaged capabilities and playbooks for agents. Install any of them with one command.
Agent workflows
Guides hands-on development of agentic AI systems—agent loops (plan → act → observe), tool and MCP schemas, multi-agent orchestration and handoffs, state/checkpointing, HITL gates, agent prompts, reliability (retries, idempotency, cancellation), observability, trajectory evaluation, tool sandboxing, injection awareness, and deployment (API, queue, durable workflows). Framework-agnostic with optional LangGraph, Deep Agents, and Cursor SDK pointers—not full framework docs. Use for agentic AI, build an agent, agent loop, tool use, MCP integration, multi-agent, agent orchestration, LangGraph, agentic workflow, AI agent developer, agent handoff, agent memory, HITL agent, evaluate agent, or agentic application—not model training (ai-engineer), ML research (ai-researcher), product strategy only (cpo-advisor), architecture whiteboard only (agent-designer, external), generic backend without agent loops (senior-software-engineer), or red-team only (ai-redteam).
Use when the user wants tool use, MCP access, HTTP or streaming API exposure, auto-function helpers, wait-for-key behavior, or optional `agently-devtools` observation, evaluation, and playground integration through Agently-native extension surfaces rather than custom wrappers first.
Use when the user wants stable structured fields, required keys, reliable machine-readable sections, or downstream-consumable output from one model request, including prompt-config-owned output contracts, `.output(...)`, field ordering, `ensure_keys`, and structured streaming.
Create, review, rewrite, or enforce a high-quality AGENTS.md (the open “README for coding agents” format). Use when the user invokes $agents-md, says AGENTS.md, agents.md, agent instructions, project agent context, nest AGENTS.md, or asks to scaffold, audit, improve, or standardize repository guidance for AI coding agents. Always include the ASD-STE100 Communication block and the Testing block (no post-hoc unit tests, prefer E2E and integration tests only when the repo already has those setups, E2E artifacts, failure modes before isolated tests) in root AGENTS.md files.
Build automated AI workflows combining multiple models and services. Patterns: batch processing, scheduled tasks, event-driven pipelines, agent loops. Tools: inference.sh CLI, bash scripting, Python SDK, webhook integration. Use for: content automation, data processing, monitoring, scheduled generation. Triggers: ai automation, workflow automation, batch processing, ai pipeline, automated content, scheduled ai, ai cron, ai batch job, automated generation, ai workflow, content at scale, automation script, ai orchestration
Build RAG (Retrieval Augmented Generation) pipelines with web search and LLMs. Tools: Tavily Search, Exa Search, Exa Answer, Claude, GPT-4, Gemini via OpenRouter. Capabilities: research, fact-checking, grounded responses, knowledge retrieval. Use for: AI agents, research assistants, fact-checkers, knowledge bases. Triggers: rag, retrieval augmented generation, grounded ai, search and answer, research agent, fact checking, knowledge retrieval, ai research, search + llm, web grounded, perplexity alternative, ai with sources, citation, research pipeline
Implement all existing tasks in an authorized batch through implement-task, in dependency order, after checking every task and requiring all user-action blockers to be resolved first. Use when the user requests automatic multi-task implementation; preserve verification/review policy and stop for new blockers or stalled corrections.
This skill helps an LLM build correct native Model Context Protocol integrations with @ax-llm/ax. Use when the user asks about AxMCPClient, MCP transports, tools, prompts, resources, subscriptions, tasks, sampling, elicitation, roots, authentication, OAuth, MCP Apps, recording/replay, or MCP integration with AxGen, AxAgent, AxFlow, chat, optimization, and AxEventRuntime.
Beads (bd) Dolt-backed issue tracker for agent task memory. Covers CLI ops, molecules, Dolt sync, Linear/Jira/GitLab. Use when tracking tasks and dependencies with the Beads CLI, syncing issues via Dolt, or integrating with Linear/Jira/GitLab. Keywords: bd, beads, Dolt, issue tracker.
Implement the Business Brain pattern. Gives every agent skill access to your tone, audience, and positioning without bloating the context window. Use when asked for brand context, voice, or positioning, or when setting up a new project's brain.
Run chief-agent across as many batches as it takes to finish a milestone, writing one report per task instead of one per batch. When a task hits ambiguity, a throwaway decision-support agent proposes options; chief-agent still makes the final call and the report captures the reasoning. Requires goals and contracts to exist. Use "/chief-loop".
(NS) Resolve ambiguities, gaps, and unstated assumptions in a scope description before generating requirements. On brownfield repos, runs bootstrap-brownfield when brownfield-map.md is missing; when a map exists, mandatory human gate to refresh or keep it (shows last map date). Use when the user gives a vague or incomplete scope, mentions "dashboard", "notifications", "integration with X" without detail, or before requirements generation when terms are ambiguous — even if they do not ask to "clarify". Do NOT generate requirements.md in this skill; output enriched context only.
Activate a 4-stage coding discipline framework that forces Claude to plan before coding, isolate changes on a branch, write tests first, and self-review output twice before presenting it. Use when starting a complex coding task, when past Claude sessions produced broken first drafts, or when you want to prevent rework cycles. Produces a confirmed written plan, isolated feature branch, test-first implementation, and a double-reviewed output with a correctness and code-quality checklist.
Guide to the Cloudflare Agents SDK for building stateful AI agents on Workers plus Durable Objects. Covers the Agent class, this.setState and this.sql state, WebSocket and streaming chat via AIChatAgent and useAgentChat, scheduling with this.schedule (delays, dates, cron), triggering Workflows, Browser Rendering, RAG with Vectorize and Workers AI embeddings, MCP servers with McpAgent, human-in-the-loop review, and routing through routeAgentRequest, getAgentByName, useAgent and AgentClient. Documents 15+ pitfalls: missing new_sqlite_classes, non-atomic migrations, Agent class not exported, binding name mismatches, global uniqueness surprises, state size limits, MCP transport confusion and bypassed authentication. Use when building AI agents, chat agents, MCP servers or multi-agent systems on Cloudflare, or debugging Agent binding, migration and state-sync errors.
Use when receiving code review feedback (especially if unclear or technically questionable), when completing tasks or major features requiring review before proceeding, or before making any completion/success claims. Covers three practices - receiving feedback with technical rigor over performative agreement, requesting reviews via code-reviewer subagent, and verification gates requiring evidence before any status claims. Essential for subagent-driven development, pull requests, and preventing false completion claims.
Inspect an existing repository and refresh only the generated section of CONTEXT.md with attributable Observed, Inferred, and Unknown evidence. Use for onboarding, material repository-context gaps, or explicit refresh; not as a mandatory stage during ongoing implementation.
MANDATORY handoff to the local Codex CLI. Triggers when (a) the user's message begins with `/codex` as a command, or (b) the user issues an explicit handoff directive like "hand this to codex", "run this through codex", "ask codex", or "have codex do X". On trigger, pipe the user's prompt verbatim to `codex exec` and return Codex's final message verbatim. Treat this like a shell alias the user is executing through you. Do NOT interpret the task, inspect files, gather context, attempt the work yourself, judge whether Codex is the right tool, or rewrite the prompt. The only abort condition is an empty prompt (ask what Codex should do).
An Instinct-based learning system that observes sessions through hooks, creates atomic instincts with confidence scores, and evolves them into Skills, Commands or Agents. Version 2.1 adds project-scoped instincts to prevent cross-project contamination.
Author, review, or refactor a skill for the SAP Fiori Tools / open-ux-tools ecosystem. Use this skill whenever the user wants to create a new skill, rewrite an existing one, add references, split an oversized SKILL.md, enforce the `sap-fiori-<feature>` naming convention, validate frontmatter, add CAP/RAP samples, declare OData V2 vs V4 requirements, mark a skill as internal (developer-facing) or external (public), or produce a verification checklist — even if the user does not explicitly say "skill". Enforces host-agnostic tooling, canonical fallbacks, explicit scope boundaries, and progressive-disclosure chunking (<500 lines per file).
Use when a new skill needs to be created or an existing skill updated, packaging domain knowledge, workflows, scripts and tools into a reusable Skill; generated new skills must be written in Chinese. Trigger words: create a new skill, update a skill, create a skill package, extend AI capabilities.
Comprehensive CrewAI framework guide for building collaborative AI agent teams and structured workflows. Use when developing multi-agent systems with CrewAI, creating autonomous AI crews, orchestrating flows, implementing agents with roles and tools, or building production-ready AI automation. Essential for developers building intelligent agent systems, task automation, and complex AI workflows.
Anime character skill distiller. Give a character name and work title: it retrieves core material locally → cross-validates → distills behavior → generates a runnable, traceable, updatable character Skill. Trigger words: "generate XX's skill", "distill XX", "make an XX character", "turn XX into a skill", "create an XX", "/csp". Vague requests also trigger: "I want to chat with a tsundere character", "any yandere recommendations", "help me make a character from XX". Requires Python; core sites use local scripts first, while web search skill / MCP / WebFetch only supplement and serve as a fallback on failure.
Executes agent-enhanced council queries as one Workflow of parallel Claude analyst agents that each query a provider, evaluate response quality, ask follow-up questions, and return a schema-enforced analysis with confidence ratings and blind spot analysis. Invoked when the --agents flag is used or when complex architectural decisions are detected.
Deep research with hyperresearch, for Claude Code and OpenAI Codex. Use when the user asks for deep research, a research report, a literature review, or a multi-source analysis with verified citations. Checks that the hyperresearch CLI is installed, sets it up in the current project for the agent you are running in, then hands off to the hyperresearch pipeline (a tier-adaptive 16-step pipeline with a persistent source vault). Not for quick lookups one or two searches can answer.
Structured development workflow that separates research, planning, and implementation into distinct phases with persistent markdown artifacts. Use when starting any non-trivial feature, refactor, bug investigation, or codebase change. Trigger on: "deep work", "research and plan", "plan before coding", "write a plan", "research this codebase", "don't code yet", "understand then implement", or when the user wants a disciplined approach to a complex task. Also use when the user says "research", "plan", "annotate", "implement the plan", or references research.md/plan.md artifacts.
The default web content reader for OpenClaw. Reads X (Twitter) tweets/profiles, Reddit, YouTube, and any webpage into clean Markdown — zero API keys required. Use when you need to ingest social media posts, profiles, articles, or video transcripts into agent memory.
Use for Desktop Commander MCP capabilities — persistent shells and REPLs, long-running processes, filesystem beyond the workspace, structured files (.xlsx, .docx, .pdf, images) and large local data files such as CSVs, ripgrep search at scale, SSH, or cross-turn state.
Autonomous harness for building complete applications from a single prompt. Uses dkod for parallel agent execution with AST-level semantic merging. Orchestrates a Planner that decomposes work by symbol into parallel units, N Generator agents that implement simultaneously via isolated dkod sessions, and a skeptical Evaluator that tests the live result via Playwright CLI (preferred) or chrome-devtools MCP (fallback), plus dk_verify. Fully autonomous — zero user interaction from prompt to working, tested PR. Use this skill whenever the user provides a build prompt ("build a...", "create a...", "make a...") or invokes /dkh.
Guide to building conversational AI voice agents on the ElevenLabs Agents Platform (ASR, LLM, TTS, turn-taking model). Covers agent configuration and system prompts, workflows, multi-voice, pronunciation dictionaries and speed control, knowledge base RAG, client/server/MCP/system tools, Scribe real-time STT, WebRTC and WebSocket connections, testing and evaluation, analytics, cost optimization, GDPR/HIPAA/SOC 2 compliance, and agents-as-code CLI workflows. Includes production templates for @elevenlabs/react, @elevenlabs/client, @elevenlabs/react-native, Next.js, Swift and Cloudflare Workers, plus 17+ documented fixes: deprecated @11labs packages, Android audio cutoff, CSP violations, missing dynamic variables, case-sensitive tool names, webhook authentication failures and WebRTC misconfiguration. Use when building, configuring or debugging an ElevenLabs voice agent.
Builds MCP (Model Context Protocol) servers in Python with FastMCP, covering tool, resource, and prompt definitions, storage backends (memory, disk, Redis, DynamoDB), server lifespans, the 8 built-in middleware types, server composition via import and mount, OAuth Proxy and authentication, OpenAPI integration, client configuration, and FastMCP Cloud deployment. Includes templates for basic servers, storage, middleware, composition, OAuth proxy, API integrations, testing, and production architectures, and prevents 25+ common errors such as storage misconfiguration, lifespan and middleware-order mistakes, circular imports, module-level server issues, async/await confusion, OAuth security holes, and cloud deployment failures. Use when creating or debugging a Python FastMCP server, defining mcp tools, resources, or prompts, wiring fastmcp storage or middleware, mounting sub-servers, setting up MCP OAuth, or deploying to FastMCP Cloud.
Helps beginners find their first AGENTIC project through pain point identification. Guides through questioning, analysis, project suggestion, and folder setup. Use when someone asks 'what project should I build?' or 'help me get started with Claude Code'.
Fortytwo Prime: collective multi-agent inference for high-stakes questions. ALWAYS use when the user says "Ask Fortytwo", "Ask Prime", "Ask Fortytwo Prime", or calls ask_fortytwo_prime. When the user doubts your answer ("try again", "that's wrong", "are you sure?"), wants more depth ("go deeper", "elaborate", "full picture"), or asks a question spanning multiple domains where a single model may fall short — suggest Fortytwo Prime but wait for confirmation before calling (this is a paid service). Even if the user doesn't mention Fortytwo by name, suggest it for contested topics, niche expertise, and high-stakes decisions.
Senior full-stack engineer mode. The agent owns technical decisions and implementation; the user is the engineering manager who sets the what/why and approves the Stage 3 plan before any code is written. Full-stack feature work in existing codebases: backend APIs, services, data models, migrations, background jobs, external integrations, CLIs, libraries, plus product UI (pages, forms, tables, modals, dashboards). Use when the user wants to implement / add / build / extend a feature: "let's build X", "add feature Y", "implement Z", multi-module refactors, schema changes, new endpoints, new pages or components in an existing product, or any non-trivial engineering task that requires reading existing patterns before writing code. Prefer over a plain coding response whenever the feature has more than one moving part.
Git + worktree orchestrator: status survey, per-worktree triage, commits, PRs, branches, releases, rebases — reads inline, writes to a background agent. Triggers on: git status, anything to commit, anything to push, commit, push, create PR, rebase, release, tag, changelog, worktree, land worktrees.
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.