Skills
96,981 packaged capabilities and playbooks for agents. Install any of them with one command.
Agent workflows
Run an autonomous Humanize-governed vLLM SOTA performance loop for one LLM model: first perform the fixed fair vLLM/SGLang/TensorRT-LLM deployment search and benchmark, then start one RLCR loop that repeatedly decides the gap, profiles the current bottleneck, runs layer/kernel pipeline analysis, patches vLLM code, optionally uses ncu-report-skill for kernel evidence, and revalidates until vLLM matches or beats the best observed framework under the same workload and SLA.
P2P communication between AI agents using walkie-sh CLI. Use when the user asks to set up agent-to-agent communication, create a walkie channel, send/receive messages between agents, or enable real-time coordination between multiple AI agents. Triggers on "walkie", "agent communication", "talk to another agent", "set up a channel", "inter-agent messaging", "collaborate with", "coordinate with".
Use for Gravity stack setup and orchestration: SatisPress/Composer configuration for the full plugin suite (Gravity Forms, SMTP, Connect, GravityView, Spellbook, Pods), reading the .cursorrules and llms.txt templates, or understanding how all products fit together. For product-specific work, load the individual skill instead.
This skill should be used when a developer wants to capture learnings from a difficult session, record what Claude got wrong, save implementation gotchas, or update the steering docs with hard-won knowledge — for example "let's reflect", "capture what we learned", "that was painful, save this", "update the steering docs with what went wrong", "I need to debrief", "what went wrong today", "log this lesson", "save this gotcha", "document this mistake", "I want to write this down before I forget", "add this to the steering docs", or when prompted by the intervention tracker after multiple corrections. Routes each learning into the right steering doc (TECH, QA, DESIGN, or VISION) under a "Hard-Won Lessons" section.
Z.AI CLI providing: - Vision: image/video analysis, OCR, UI-to-code, error diagnosis (GLM-4.6V) - Search: real-time web search with domain/recency filtering - Reader: web page to markdown extraction - Repo: GitHub code search and reading via ZRead - Tools: MCP tool discovery and raw calls - Code: TypeScript tool chaining Use for visual content analysis, web search, page reading, or GitHub exploration. Requires Z_AI_API_KEY.
Assesses and improves a git repository's readiness for AI coding agents with the agentready tool, and reports the score, the certification level, the failing findings, and where the report was written. Covers one repository or every RHDH repository under a directory, and applies the fixes the report supports. Use for "assess agent readiness", "run agentready", "improve our agent readiness score", "prepare this repository for coding agents", or "assess all the RHDH repositories".
A security testing methodology for AI Agent systems. Triggers when the target system uses AI Agents for tool invocation, multi-agent collaboration, or autonomous decision-making. Covers the OWASP Agentic AI Security Top 10 (ASI01-ASI10): goal hijacking, tool misuse, identity and privileges, supply chain, code execution, memory poisoning, multi-agent communication, cascading failures, human-agent trust exploitation, and rogue agents.
LOAD THIS SKILL when: using CLI wrapper tools (gh-tool, observability-tool, db-tool, k8s-tool, az-tool, azdo-tool, logs-tool, session-tool, api-tool, api-proxy), working with observability, databases, GitHub PRs, Kubernetes, Azure platform resources, Azure DevOps, or application logs. Contains tool overview, usage patterns, and project-specific aliases.
Generate AGENTS.md and CLAUDE.md files for a repository. AGENTS.md provides cross-tool agent instructions (supported by Claude Code, Cursor, Windsurf, Zed, Codex, and others). CLAUDE.md adds Claude-specific configuration and references AGENTS.md via @import. Use when a repo needs agent onboarding or when starting a new project.
Full delivery pipeline: plan, build, ship, settle. Covers: shape/spec/design, TDD build, commit, PR creation, PR fix (CI/reviews/conflicts), PR polish, simplify, test coverage, verify ACs, walkthrough, issue management. Use when: shipping features, fixing PRs, creating PRs, building issues, simplifying code, checking quality, writing commits, managing issues. Trigger: /autopilot, /build, /shape, /commit, /issue, /check-quality, /test-coverage, /verify-ac, /pr-walkthrough.
AWS Bedrock AgentCore comprehensive expert for deploying and managing AI agents at scale. Use when working with any AgentCore service including Gateway, Runtime, Memory, Identity, Code Interpreter, Browser, Observability, Agent Registry, or Evaluations. Covers agent deployment, MCP tool integration, credential management, agent discovery, governance workflows, and automated quality assessment. Essential when user mentions AgentCore, agent runtime, agent registry, agent evaluation, MCP gateway, deploy agent, register MCP server, discover agents, evaluate agent quality, agent credentials, or wants to build, deploy, catalog, or monitor AI agents on AWS.
Amazon Bedrock AgentCore platform for building, deploying, and operating production AI agents. Covers Runtime, Gateway, Browser, Code Interpreter, and Identity services. Use when building Bedrock agents, deploying AI agents to production, or integrating with AgentCore services.
ALWAYS use this skill before answering brainstorming, ideation, prompt crafting, or open-ended exploration requests. Transforms vague requests into actionable outputs via adaptive guided questioning — triages into Prompt Mode (craft/improve prompts), Explore Mode (brainstorm ideas), or Focused Mode (specific problem strategies). Trigger when user says: "brainstorm", "ช่วยคิด", "help me think", "I have an idea", "improve this prompt", "let's explore", "I want to build", "I'm thinking about", "brainstorm วิธี", "ช่วยคิดหน่อย", "อยากทำ", "ยังไม่รู้จะทำอะไร", "not sure about the approach", "help me figure out", "what should I". Also trigger for: side projects, career decisions, project planning, migration strategies, architecture decisions, cost optimization, or any request where the user hasn't decided direction yet and would benefit from structured discovery. Do NOT skip — adapts depth automatically (2-7 questions) and produces BETTER results by asking targeted questions first.
Build a Loop for PostHog Desktop: a workflow that creates an AI task each time its trigger fires, optionally followed by a Slack or email notification with the task's result. Use when asked to create, set up, or change a loop, a recurring agent, a scheduled task, or an automation that runs an AI task when a GitHub, Slack, or PostHog event happens. Covers the exact graph, trigger configs, schedule presets, the notify step, structured output, and the test-run steps.
Use when acquiring unfamiliar business knowledge and distilling it into a BK-CI skill or architecture document, such as reading iWiki, cross-validating with code, extracting module boundaries, and rewriting knowledge documents. Prefer this when the user wants to understand the business before writing documentation.
Decision guide for delegating to caveman-style subagents. Tells the main thread WHEN to spawn cavecrew-investigator (locate code), cavecrew-builder (1-2 file edit), or cavecrew-reviewer (diff review) instead of doing the work inline or using vanilla Explore. Subagent output is caveman-compressed so the tool-result injected back into main context is ~60% smaller — main context lasts longer across long sessions. Trigger: "delegate to subagent", "use cavecrew", "spawn investigator/builder/reviewer", "save context", "compressed agent output"
[Hyper] Use when the user explicitly wants Anthropic Claude Code CLI (`claude`) for an isolated session, non-interactive (`-p`) run, session resume by ID or name, or a CI-friendly `--bare` invocation. Trigger phrases: "use claude code", "ask claude", "run claude", "continue the last claude session", "resume the auth-refactor claude session", or "use Anthropic's CLI to inspect or fix this repo".
Built-in memory decides what to remember. This one asks. Agent memory as plain files in your folder: the agent proposes, you approve, every line dated. One memory per client, session handoffs, promotion into knowledge and rules on your yes, plus orchestration and QA layers.
Evaluate Clojure skills through rigorous real-project testing with and without skills enabled. Use this skill whenever you need to assess whether a Clojure skill (like clj-discover, clj-debug, clj-replace) actually improves development outcomes. Compare behavior differences via context-logs, code quality, documentation depth, and test comprehensiveness. This prevents vague or hypothetical evaluation and forces concrete real-world testing.
Exports a Claude.ai workspace — its custom instructions and its uploaded knowledge files — into a bundle that can be re-imported elsewhere. Use for "back up my project instructions", "export my project knowledge", "copy this workspace to a new one", "migrate my project to another account", or "recreate this setup somewhere else". Operates on the Claude.ai workspace you are talking in, and reads nothing from disk
The reference merchant agent, covering its flows, staged changes and host approval, metrics grounding, the analysis delegate, store memory, components, and marketplace rules. Load when building or reviewing an agent for the operators of a store, property, subscriber base, or venue.
Only use when the user explicitly invokes /create-skill (or $create-skill in Codex). Never auto-invoke. A skill for interactively creating a new Agent Skill from scratch. Use it when the user says they want to create a skill or turn a task into a skill. Ask the user questions to clarify intent (the problem to solve, the situations that should trigger it, the scope of use), then consistently carry out the decision on whether it should become a skill, the choice of layer (shared layer / project layer) and name, the design and writing of SKILL.md, and verification with a checker. The source of truth for general skill-writing principles (progressive disclosure, writing descriptions, etc.) is agent-native-project-design; this skill provides the creation workflow that applies them.
This skill should be used when the user asks to "create a skill" or "make a command". Make sure to use this skill whenever the user mentions skill creation, command authoring, slash commands, or building Claude extensions — even if they don't explicitly say "create-skill". Not for repairing or auditing existing skills — use repair-skill.
Builds and revises a complete skill DIRECTORY — SKILL.md, scripts, references, assets — and packages it. Use for "create a skill for X", "turn this into a skill", "update/improve this skill", "why doesn't my skill trigger", "review this SKILL.md", "package this skill", or when a repeated procedure should become a reusable artifact. Enforces the structure that makes skills work — a concrete procedure rather than a goal statement, an explicit applicability boundary, failure modes with their signals, a runtime verification step, and a description checked against the existing catalogue for confusability. For choosing whether the instruction should be a skill at all rather than project instructions or a prompt, use crafting-instructions. For writing quality inside the prose, use writing-instructions
Multi-agent orchestration framework for autonomous AI collaboration. Use when building teams of specialized agents working together on complex tasks, when you need role-based agent collaboration with memory, or for production workflows requiring sequential/hierarchical execution. Built without LangChain dependencies for lean, fast execution
Multiagent AI system for scientific research assistance that automates research workflows from data analysis to publication. This skill should be used when generating research ideas from datasets, developing research methodologies, executing computational experiments, performing literature searches, or generating publication-ready papers in LaTeX format. Supports end-to-end research pipelines with customizable agent orchestration.
Terminal/PTY automation with persistent sessions. Use to run and interact with TUI applications, debug terminal apps, automate CLI workflows, or any terminal interaction. Trigger phrases include "run the TUI", "start the app", "debug the terminal", "interact with", "send keys", "what's on screen".
Use first when Arduino or embedded hardware work spans sessions, parts delivery, wiring, soldering, flashing, measurements, field tests, recovery, or a durable next-todo is needed. Maintain explicit evidence and concrete physical-world user gates instead of assuming completion.
Work an entire epic of tickets unattended — for each ticket: plan it, implement it, PR it into an epic branch, drive CI green, merge; keep the epic branch synced with its target; then verify the assembled epic once, open an epic PR that merges cleanly, and file follow-ups. Works with any tracker: GitHub, Jira, GitLab, or a markdown checklist. Use when the user hands over a batch of related issues and wants them worked through autonomously ("run this epic", "work through these tickets", "do the whole milestone"). For a single ticket where the work deserves an adversarial implement/review loop, use `player-coach` instead.
Create and run LangWatch experiments for pre-deployment batch testing. Use when the user wants to test an agent against a dataset, compare prompts or models, benchmark quality, detect regressions, or add a CI quality gate. Do not use for production monitoring or guardrails.
Helps users discover and install agent skills when they ask questions like "how do I do X", "find a skill for X", "is there a skill that can...", or express interest in extending capabilities. This skill should be used when the user is looking for functionality that might exist as an installable skill.
Orchestrate the path from a raw research idea to a formal claim with a defensible proof design — drawing the hypothesis out of the idea, stating it as a precise theorem/claim, and designing (and auditing) the proof strategy. Use whenever the user says "help me turn this idea into a theorem", "I have a hunch, can we make it rigorous", "design a proof for this", "what would I need to prove this", "formalize this claim", "is this provable and how", or hands over an informal conjecture they want to develop into a formal result. This is a ROUTER that sequences existing singleton skills; it does not write the proof itself. Use flow-learn-new-topic or flow-deep-understanding when the goal is to understand existing work rather than formalize the user's own idea. Use gap-finder when the starting point is an existing paper's weakness rather than a fresh idea.
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.