Skills
97,501 packaged capabilities and playbooks for agents. Install any of them with one command.
Agent workflows
Create and configure AiderDesk agent profiles by defining tool groups, approval rules, system prompts, subagent settings, subagent filtering, and provider/model selection. Use when setting up a new agent, creating a profile, or configuring agent tools, permissions, and subagent behavior.
Transition from static LLM chats to autonomous agents that execute multi-step tasks. Use this when you need to automate cross-platform reports (e.g., Snowflake to Google Docs), build self-service tools for non-technical teams, or create "anticipatory" engineering workflows that draft PRs based on Slack discussions.
Transform Claude Code into an AI Scientist that orchestrates research workflows using tree-based hypothesis exploration. Triggers on "research project", "scientific experiment", "run experiments", "AI scientist", "tree search experimentation", "systematic study".
Turn an ambitious or unclear idea into a workable direction and carry the initiative through its requested outcome. Use when the user does not know where to start, needs alternative directions, faces dependent unknowns, or needs planning, coordinated delivery, a behavior contract or decomposition. Settled coding changes belong to alaga.
Interactive escalation-resolution workflow for the bmad-loop orchestrator. A bmad-loop run paused on a CRITICAL escalation (a contradiction or gap a dev/review session could not safely resolve alone); you and the human disambiguate the frozen spec so the story can be re-driven. Invoked as /bmad-loop-resolve <story-key>. Unlike the automated dev/review sessions this session is interactive — a human is present and you SHOULD ask.
Triage the deferred-work ledger for the bmad-loop orchestrator: verify every selected open entry against the actual codebase and return a machine-readable partition (bundles, already-resolved, blocked, skip, human decisions). Also migrates legacy pre-DW-format ledgers when invoked with --migrate. Automation-only — invoked by bmad-loop sweep runs, not by humans.
Explore user intent, requirements, and design options through collaborative dialogue before implementation. Use before building new features, components, or systems — whenever the user describes something to build and design decisions are involved. Triggers: "brainstorm", "help me design", "think through the requirements", "头脑风暴", "设计方案", "梳理需求". Not for bug fixes, config changes, or tasks with an obvious implementation path.
Close out a finished <DOC>.final.md — fold what landed into the docs that already exist, carry what is still owed into <DOC>-V2.rev1.md, and delete the planning artifacts the tree no longer needs. Use when the user invokes /close-plan <feature>, or asks to clean up or retire a small feature's docs after execute-plan or build-feature has finished. Reviews ACCEPTANCE.md check by check before calling anything landed, then triages every phase as landed, pending, deferred or dropped. The smaller sibling of close-roadmap — works for any stem (PLAN, TECHNICAL, …); a plan usually does not earn its own folder of permanent documentation, so the default is to add a paragraph to a page that exists rather than create a new one
Close out a finished roadmap or feature execution by retiring its planning docs into canonical documentation and carrying what is still owed into the next roadmap — you are the archivist and the architect of the next round, not the coder. Use when the user invokes /close-roadmap <feature>, or asks to clean up a feature's docs, retire a roadmap, prepare for the next roadmap, or tidy docs/<feature>/ after execute-roadmap or build-feature has finished. Reviews ACCEPTANCE.md check by check, triages every item of ROADMAP.final.md or PLAN.final.md as landed, pending, deferred or dropped; moves pending work into ROADMAP-V2.rev1.md; turns landed work into decision records, architecture, data-model and product docs; writes the folder's README.md; deletes revs and execution artifacts the tree no longer needs; and fixes every inbound link. Nothing disappears unaccounted for; the user settles anything that would be dropped. For a small plan whose landed work folds into pages that already exist, use close-plan instead
This skill should be used when creating a Claude Code slash command. Use when users ask to "create a command", "make a slash command", "add a command", or want to document a workflow as a reusable command. Essential for creating optimized, agent-executable slash commands with proper structure and best practices.
Plan a large feature as a ROADMAP.rev1 written with the user — you are the product manager first and the senior software architect second, never the coder. Use when the user asks to plan a large feature, write a roadmap, or invokes /draft-roadmap <feature>. Settles the product shape — the one sentence, and the ladder of use cases smallest to largest — then derives a roadmap of small, verifiable, delegatable milestones and plans from it, ready for review and for parallel execution
Use this skill when the user's Copilot Studio agent evaluations have come back and they need to interpret scores, diagnose root causes of underperforming test cases, find remediation steps, or analyze patterns to improve their agent. Always use this skill when the user mentions: "eval failed", "why did this fail", "triage", "diagnose failure", "low pass rate", "fix evaluation results", "not passing", "failing test cases", "evaluation results", "improve my eval scores", or any situation where eval scores need interpretation and action.
Execute a signed-off PLAN.final.md for a single feature yourself, phase by phase, on the current branch — you are the engineer, not a delivery lead. Use when the user invokes /execute-plan, or asks to implement, build or continue an approved plan that is small enough for one person, e.g. "/execute-plan spell-check", "/execute-plan spell-check p1-p3", "build the rest of the plan while I sleep". Resolves the requested phases against the final doc, builds each as its own verified increment, runs its verification before the next begins, and commits each landed phase without waiting for approval. Questions that need the user are recorded with a recommendation instead of blocking. For a roadmap large enough to want parallel sub-agents, use execute-roadmap instead
Execute an approved roadmap or plan by orchestrating sub-agents — in parallel, in isolated checkouts, where the runtime allows — you are the delivery lead, not the coder. Use when the user invokes /execute-roadmap, or asks to implement, execute, or oversee an approved roadmap or plan after draft-roadmap or build-feature has produced a signed-off final doc, e.g. "/execute-roadmap implement m2, plans 3.1-3.4, parallelize" or "oversee the roadmap until it is complete". Resolves the requested scope against the final doc, gives each plan to its own sub-agent, reviews and verifies each result against the project's principles, then merges, cleans up, and commits without waiting for approval. Honors model/effort directives where the agent tool supports them; questions that need the user are recorded with a recommendation instead of blocking.
Workspaces for agentic teams. Complete agent guide with all 19 consolidated tools using action-based routing — parameters, workflows, ID formats, and constraints. Use this skill when agents need shared workspaces to collaborate with other agents and humans, create branded shares (Send/Receive/Exchange), or query documents using built-in AI. Supports ownership transfer to humans, workspace management, workflow primitives (tasks, worklogs, approvals, todos), and real-time collaboration. Free agent plan with 50 GB storage and 5,000 monthly credits.
Create and synchronize prompt-based AI agents directly within Azure AI Foundry via REST API, from a local JSON manifest. Unlike scaffolding skills that only generate local code, this skill registers agents in the Foundry service itself — making them immediately available for invocation. Use when the user asks to create agents in Foundry, sync, deploy, register, or push agents to Foundry, update agent instructions, or scaffold the manifest and sync script for a new repository. Triggers: 'create agent in foundry', 'sync foundry agents', 'deploy agents to foundry', 'register agents in foundry', 'push agents', 'create foundry agent manifest', 'scaffold agent sync'.
Best practices for building AI agents with Google's Agent Development Kit (ADK) in Python, covering agent design, tools, sessions, memory, artifacts, evaluation, and deployment. Use when building LLM agents or multi-agent systems with ADK, defining ADK tools, wiring up sessions/state/memory, working with ADK artifacts, writing agent evals, or deploying ADK agents to Vertex AI Agent Engine or Cloud Run.
Create and refine HEARTBEAT.md files for murmur — a CLI daemon that runs scheduled Claude prompts on a cron or interval schedule. Use this skill when the user wants to set up a recurring automated action (e.g., "monitor my GitHub issues", "check Hacker News for AI articles", "watch my endpoints", "send me a daily digest"). Guides the user through an interview, drafts the heartbeat prompt, tests it, and registers it with murmur's scheduler. Triggers: heartbeat, murmur, recurring task, scheduled action, cron, monitor, watch, automate, periodic check, scheduled prompt.
General-purpose digital immortality framework: distills a digital twin of anyone from multi-platform data such as chat logs, social media and documents — supports 7 role templates (yourself, colleague, mentor, family member, partner/ex, friend, public figure) and connects to 12+ data platforms at home and abroad.
Browse, group, relate, and selectively install the jeo-skills catalog through the lightweight `jeo-skill` CLI. Use when the user wants skills organized by web, infrastructure, game, creative media, CLI tools, AI/agents, engineering, research, business, or utilities; needs a frontend/backend/game-audio/game-VFX subcategory; wants overlapping skills connected instead of duplicated; or wants a category, bundle, or named skills installed without copying the full repository.
Main routing skill for the lyt cross-border e-commerce series. Use when the user invokes /lyt, $lyt, Lyt, or asks about cross-border e-commerce, TikTok Shop, product selection from scratch, product validation, product title generation and optimization, product main image and detail page diagnosis, generating detail pages from product images, summarizing competitor/Top main image samples, reverse-engineering e-commerce reference image prompts and swapping in your own product, generating short video shooting scripts from existing products, follow-up questions on business problems, traffic logic, data analysis, systematic learning, or knowledge checks. First determine the business boundary and the task the user actually wants to complete, then route to lyt-product-selection, lyt-product-validation, lyt-product-title, lyt-page-diagnosis, lyt-detail-page, lyt-image-patterns, lyt-image-prompt, lyt-video-script, lyt-problem-clarifier, lyt-traffic-logic, lyt-data-analysis, lyt-learning or lyt-knowledge-check. Supports text, images/screenshots, CSV/XLSX spreadsheets, PDF/documents and backend exports; the attachment type only describes the input form and does not replace the user's intent.
Use when the user explicitly asks for Nerd Loop or wants a cost-proportional, behavior-aware task-completion controller that selects the cheapest adequate loop profile and iterates toward an explicit Definition of Done through focused work, automatic verification, and bounded stopping rules.
Multi-agent parallel development cycle with requirement analysis, exploration planning, code development, and validation. Orchestration runs inline in main flow (no separate orchestrator agent). Supports continuous iteration with markdown progress documentation. Triggers on "parallel-dev-cycle".
Set up a structured project memory for any software project — TODO.md as priority/effort-tagged index of future work, backlog/ for resume-friendly research/design notes on P? items, and pitfalls/ as a symptom-grep-able knowledge base of past traps. Use when a user wants somewhere to record "maybe later" ideas, freeze troubleshooting state, capture trade-off analysis, or stop re-debugging the same problem.
Expert prompt engineering skill that transforms Claude into "Alpha-Prompt" - a master prompt engineer who collaboratively crafts high-quality prompts through flexible dialogue. Activates when user asks to "optimize prompt", "improve system instruction", "enhance AI instruction", or mentions prompt engineering tasks.
Pick up unfinished work from docs/<feature>/HANDOFF.md in a fresh session — you are the agent the handoff was written for, and you verify it before you trust it. Use when the user invokes /resume-handoff <feature>, points at a HANDOFF.md, or asks to continue, resume, or pick up where the last session stopped on a plan or roadmap built by execute-plan, execute-roadmap or build-feature, e.g. "pick up the handoff and keep going while I sleep". Reads the project's agent orientation and the handoff, reconciles its claims with git and the code, surveys the pending merge and proposes the scope — then stops and asks the user to confirm the merge and plan of work, or give new instructions, before anything changes the tree. On confirmation it deletes the consumed handoff, merges through the merge-branch gate, continues under execute-plan or execute-roadmap, and writes a fresh handoff at the end if work is still owed. For writing a handoff, use write-handoff; for retiring finished work, use close-plan or close-roadmap.
Find, inspect, recover, identify, or resume earlier coding-agent sessions across Claude Code, Codex, and Pi histories. Use whenever the user asks to locate previous work or continue an earlier agent conversation, regardless of the currently running harness.
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.