Skills
96,981 packaged capabilities and playbooks for agents. Install any of them with one command.
Agent workflows
Persistent memory architecture for AI agents across sessions. Episodic memory (past events), procedural memory (learned skills), semantic memory (knowledge graph), short-term memory (active context). Use when implementing cross-session persistence, skill learning, context preservation, personalization, or building truly adaptive AI systems with long-term memory.
Quickly creates new Claude Code skills or translates ChatGPT projects into Claude Code skills. Handles skill scaffolding, frontmatter, directory structure, and ChatGPT-to-Claude migration. Use when the user wants to 'create a skill,' 'make a new slash command,' 'convert a ChatGPT project,' 'translate a GPT to Claude,' or 'migrate prompts to Claude Code.' For full eval/testing/benchmarking workflows, use skill-creator instead.
Use for current official documentation about Codex, Codex CLI, or Claude Code behavior, configuration, prompting, skills, permissions, tools, surfaces, capabilities, troubleshooting, hooks, app-server, or hook trust. Consult the relevant official source before answering.
Only use when the user explicitly invokes /agents-md-sync (or $agents-md-sync in Codex). Never auto-invoke. This skill analyzes the entire repository and generates/updates a hierarchical AGENTS.md knowledge base (root and subdirectories selected by scoring). Each run detects divergence between the existing AGENTS.md and the codebase and keeps it up to date, and also maintains the CLAUDE.md bridge for Claude Code. On first creation, it fills in normative information that cannot be derived from code (purpose, development phase, operational constraints) through interviews. Use it when asked to set up, initialize, or update AGENTS.md, or to build a knowledge base equivalent to init-deep. For design principles on what to write in instruction files, see agent-native-project-design.
Proactively use AgentLoop Recall to retrieve prior Alibaba Cloud AgentLoop experience through the bundled SearchContext CLI whenever the user asks or implies that prior work may help. Trigger for requests to check, search, recall, retrieve, look up, review, consult, reference, or compare prior experience, historical troubleshooting cases, past fixes, comparable incidents, lessons learned, old runbooks, previous remediations, or successful workflows before or during work. Also trigger for Chinese requests such as 先查历史经验、回忆类似案例、召回过往排障记录、看看之前有没有处理过类似问题、参考以前怎么处理、找找之前的踩坑记录、翻一下历史排障、有没有类似经验、先看看过去的案例.
Use this skill when you need to control a Chrome browser via CDP (Chrome DevTools Protocol) to reuse existing login sessions. Covers: launching Chrome in debug mode, opening URLs, waiting for page load, evaluating JavaScript, taking snapshots, and extracting auth tokens. Trigger phrases: browser automation, CDP, agent-browser, 浏览器操作, 操作浏览器, Chrome CDP, 复用登录态, extract token from browser.
Use the Browser Use cloud API to spin up cloud browsers for Clawdbot and run autonomous browser tasks. Primarily, create browser sessions with profiles (persisted logins/cookies) that Clawdbot can control. Secondarily, run task subagents for fast autonomous browser automation. Docs at docs.browser-use.com and docs.cloud.browser-use.com.
End-to-end development lifecycle for AI coding agents — scaffold projects, plan and build features, create Linear tickets, open and review PRs, clean up after merge, audit codebases, and improve UI polish/motion quality. Works with any agent that reads CLAUDE.md or AGENTS.md. Use when starting a project, implementing a feature, shipping code, addressing PR reviews, merging, auditing code quality, designing or reviewing UI polish, or reviewing animations.
Generate progressive disclosure indexes for GitHub repositories to use as Claude project knowledge. Use when setting up projects referencing external documentation, creating searchable indexes of technical blogs or knowledge bases, combining multiple repos into one index, or when user mentions "index", "github repo", "project knowledge", or "documentation reference".
Produce a long-form, shareable markdown writeup on whether Claude has regressed on this user's work. A bundled Python script scans `~/.claude/projects/`, computes every metric, and renders a markdown skeleton with tables already filled — in ~2.5s. Claude fills a dozen short narrative placeholders and saves. Writes `./cc-canary-<YYYY-MM-DD>.md` suitable for pasting into a GitHub issue or gist.
This skill provides comprehensive knowledge for working with the Anthropic Claude Agent SDK. It should be used when building autonomous AI agents, creating multi-step reasoning workflows, orchestrating specialized subagents, integrating custom tools and MCP servers, or implementing production-ready agentic systems with Claude Code's capabilities. Use when building coding agents, SRE systems, security auditors, incident responders, code review bots, or any autonomous system that requires programmatic interaction with Claude Code CLI, persistent sessions, tool orchestration, and fine-grained permission control. Keywords: claude agent sdk, @anthropic-ai/claude-agent-sdk, query(), createSdkMcpServer, AgentDefinition, tool(), claude subagents, mcp servers, autonomous agents, agentic loops, session management, permissionMode, canUseTool, multi-agent orchestration, settingSources, CLI not found, context length exceeded
Create, audit, and maintain CLAUDE.md documentation files that configure Claude Code for projects. Use this skill when (1) initializing a new project with Claude Code configuration, (2) reviewing or improving existing CLAUDE.md files, (3) organizing project instructions using progressive disclosure patterns, (4) converting repeated instructions into permanent documentation, or (5) setting up agent_docs/ structures for larger codebases. Handles the WHAT/WHY/HOW framework, conciseness optimization, and file import patterns.
Executes council queries by running the query pipeline across selected AI providers (Gemini, OpenAI, Grok, Perplexity), displaying formatted responses verbatim, and generating a synthesis of consensus, divergence, and recommendations. Invoked by the ask command during standard (non-agent) council queries.
Scaffold a new integration-type "seed skill" — a SKILL.md with reference assets and prompts.jsonl entries that teach an agent to generate working, testable integration code from scratch. - Use when: developer wants to create a new skill for an SDK or framework integration that agents should be able to implement and test - Ask first: Is this an integration skill (code generation + tests) or a tool skill (CLI wrapper with no tests)? Only proceed with this skill for integration-type skills.
Overview and guidance hub for .NET testing fundamentals. Triggers when the user asks about general testing needs such as "how to write .NET tests", "getting started with .NET testing", "which testing tools do I need", "testing best practices", or "learning testing from scratch". Recommends suitable sub-skill combinations based on the specific need, covering 19 fundamental skills including testing basics, test data, assertions, mocking, and special scenarios. Make sure to use this skill whenever the user asks about .NET testing in general, wants to know which testing tool to use, or needs guidance on getting started with testing, even if they don't explicitly ask for an overview. Keywords: dotnet testing, .NET testing, testing basics, how to write tests, testing best practices, unit test, xunit, 3A pattern, FIRST principles, assertion, mock, stub, NSubstitute, test data, AutoFixture, Bogus, validator, FluentValidation, TimeProvider, IFileSystem, code coverage, ITestOutputHelper, test naming
Design an auditable playbook when no narrower one fits: a large migration, an ambitious multi-part change, or work a human reviews after stepping away. Scale rigor to the task, run a hypothesis loop, and log decisions. Use for /figure-it-out or "figure it out".
Helps users discover and install Claude Code skills. Use this skill when the user asks "how do I do X", "find a skill that can do X", "is there a skill that can...", or wants to extend AI capabilities. Searches the skills.sh ecosystem, verifies quality, and assists with installation.
Orchestrate a research paper end-to-end — from raw idea through literature grounding, experiment design, implementation, results, writing, submission, rebuttal, and artifact release. Use whenever the user wants to drive a whole paper rather than one stage: "help me write this paper from scratch", "where am I in the paper process", "what should I do next on this project", "take this from idea to submission", "I have results, get me to a draft", "manage my paper workflow", or hands over a project and asks for the path to a venue. Also use it for the writing phase itself — "draft the introduction", "outline this paper", "how should I structure this", "turn this spine into a draft", "my first paper" — which it owns as a phase-gated stage-group (stage 10) rather than delegating. This is otherwise a ROUTER that sequences existing singleton skills with stage gates; it does not do the ideation or experiments itself. For a single narrow section, invoke that section's skill directly (e.g. abstract-and-intro-writer, method-section-writer, results-writeup). For PhD application packages use flow-phd-application. For talks/decks use the presentation skills.
Execute prompts as the Gemma seat through the Pi CLI on OpenRouter, with google/gemma-4-31b-it pinned per invocation. Use when users explicitly request Gemma execution, when a workflow needs a Gemma seat, or when a cross-runner workflow wants a Gemma-backed perspective without leaving the current workspace.
Goal-driven long-run workflow: run an agent autonomously for hours toward a verifiable goal. Five stages — ① deep intake interview and output contract (approach frozen before launch) ② layered acceptance criteria + design the /goal condition (measurable end state, stated check, constraints, turn/time cap) ③ sub-agent division and context management (context-rot mitigation) ④ launch the long run (/goal, claude -p non-interactive, or manual-loop fallback) ⑤ completion report and human acceptance. Built on Claude Code’s native /goal harness, with a generic fallback for environments without /goal. Triggers — 「goal 长跑」「goal run」「goal-driven」「目标驱动长跑」「一个 goal 下去跑」「长跑目标」「无人值守跑任务」「goal-run」 / goal run, goal-driven, long-run goal, autonomous run, run until done
Constrains LLM output during generation with Guidance (Microsoft Research), using regex, select() choices, context-free grammars, token healing, and @guidance functions, with Anthropic, OpenAI, Transformers, and llama.cpp backends. Use when you need generated text to match a regex or fixed format such as dates, emails, or IDs. Use when you need guaranteed valid JSON, XML, or code from a model. Use when building multi-step generation workflows or ReAct-style agents with Python control flow. Use when classifying text into fixed categories with select(). Use when running local models and wanting grammar constraints. Not for Pydantic validation with automatic retries; use Instructor for that.
Continuous self-improvement system for Hermes Agent. Analyzes your past sessions to find recurring failures and skill gaps, then automatically creates or patches skills and runs self-evolution to fix them. Set it to run overnight and wake up to a better agent. Use /dojo to start.
Create event-driven hooks for Claude Code automation. Use when the user wants to create hooks, automate tool validation, add pre/post processing, enforce security policies, or configure settings.json hooks. Triggers: create hook, build hook, PreToolUse, PostToolUse, event automation, tool validation, security hook
This skill should be used when the user asks to "improve a skill", "make this skill better", "add features to a skill", "this skill is missing something", "upgrade my skill", "what's missing from this skill", "the skill doesn't do X", "make this more useful", or wants to improve skill effectiveness rather than structural correctness. Not for structural fixes — use repair-skill. Not for agents.
Turns friction from an Infrahub skill session into a reviewed GitHub issue about an Infrahub skill's guidance. Infrahub skills fail quietly: a missing or unclear rule does not crash, it produces extra round trips and repeated user nudges. This skill gathers evidence of that friction, checks the tracker for an existing report, works out whether the skill or the underlying product is at fault, and drafts a proposed rule change. It never files anything itself: it hands the redacted draft to infrahub-reporting-issues, which shows it to the user and gets explicit approval before any submission. TRIGGER when: accepting a friction offer, the user says "report skill friction", asks why an Infrahub skill kept failing or needed so many retries, or wants a gap in an Infrahub skill's guidance reported. DO NOT TRIGGER when: reporting a bug in Infrahub itself or any other opsmill/infrahub-* product (use infrahub-reporting-issues instead), or for ordinary Infrahub work where nothing went wrong.
Multi-conversation methodology for iterative stateful work with context accumulation. Use when users request work that spans multiple sessions (research, debugging, refactoring, feature development), need to build on past progress, explicitly mention iterative work, work logs, project knowledge, or cross-conversation learning.
Turn complex or visual agent responses into rich, reviewable HTML artifacts the user can annotate and send feedback on, using the lavish-axi CLI. Use when about to give a plan, comparison, diagram, table, code diff, report, or anything easier to grasp visually than as prose.
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.