Skills
97,725 packaged capabilities and playbooks for agents. Install any of them with one command.
Agent workflows
Captures real usage feedback for agile skills and turns it into evidence-backed proposals to improve, merge, split, deprecate, or remove skills and templates. Use when a skill does not fit a real workflow, when templates need refinement, when skills overlap, or when the process library needs maintenance.
Begin a server-planned Altitude journey and bind it to a local workshop. Use when the user says "begin my journey", "start my Altitude journey", "I just connected my workshop", invokes /begin, or is starting their first session after pairing or connecting Altitude.
Trigger this skill immediately when the user says "bootstrap" in any form. Also trigger it for any request about initializing a project's workflow, configuring development practices, or setting up a codebase for the first time. Key signals: "bootstrap", "set up workflow", "configure TDD", "generate AGENTS.md", "recommend skills", "onboard to project", "what tools should I use", "greenfield", "just cloned", "new developer joining". Detects the project environment, recommends skills by phase, configures TDD strategy, and generates instruction files. Project-level onboarding -- NOT for writing code, tests, or performing specific dev tasks.
Hosted-gateway LLM router — save 84% on inference costs. A local proxy that forwards each request to the blockrun.ai gateway, which routes to the cheapest capable model across 90 models from OpenAI, Anthropic, Google, DeepSeek, xAI, Z.AI, and more. Free open-weight models included. Also exposes realtime market data (global stocks, crypto, FX, commodities), Twitter/X intelligence, prediction-market data across Polymarket, Kalshi, Limitless, Opinion, Predict.Fun, dFlow + UMA oracle resolution + wallet identity & clustering, phone-number intelligence (carrier + SIM-swap fraud detection) plus AI-powered outbound voice calls (Twilio + Bland.ai) as built-in agent tools. Not a local-inference tool — prompts are sent to the blockrun.ai gateway.
CodeStable: understand and drive software evolution with Vision / Project Spec / Epic / Issue and codestable/ institutional memory; act according to the user's current stance—discussion, quick fixes, managed implementation, bug fixing, organizing vision/spec, understanding the status quo, closing out, or review as needed. Triggers: cs, CodeStable, discussion/talk it through first, quick/quick fix/dive right in, spike, fix bug/debug, organize vision/spec, close/wrap up, review, how does this system work, or the project already has codestable/ and is working on vision, spec, bug, implementation, and closing.
Orchestrate ledger-tracked work through subagents. Use for a bundled GitHub ticket with 3+ sub-issues, a request for subagent-driven execution, resuming such a run after compaction, or explicit away/keep-going delegation even when only one task remains.
Guide an iterative product and technical discussion, then turn the settled feature into a copy-ready orchestration prompt for a separate implementation session with a user-chosen model roster for its Implementer, Researcher, and Reviewer sub-agents. Use when the user wants to explore, narrow, summarize, or hand off a substantial feature without implementing it in the current session.
Fetch GitHub issues, spawn sub-agents to implement fixes and open PRs, then monitor and address PR review comments. Usage: /gh-issues [owner/repo] [--label bug] [--limit 5] [--milestone v1.0] [--assignee @me] [--fork user/repo] [--watch] [--interval 5] [--reviews-only] [--cron] [--dry-run] [--model glm-5] [--notify-channel -1002381931352]
Docs-anchored grilling session — challenges a plan against the project's existing language (CONTEXT.md) and recorded decisions (docs/adr/), and updates those files inline as terminology and decisions crystallise. Use when user wants to stress-test a plan against documented domain language, or mentions "grill with docs".
Create a structured session handoff document for continuity across sessions. Use when ending a work session, switching contexts, or before a break. Captures decisions, progress, code changes, and next steps so a future session can pick up where you left off without losing context.
Make every reply use natural Taiwan Traditional Chinese and open with the answer, the completed result, or the genuinely needed next step. Use this skill when replying to any user message, including programming, debugging, explanations, planning, research, and everyday conversation. When the question already provides the cause, answer only that cause and the direct fix, without adding diagnostic checklists or unverified assumptions. Keep code, commands, paths, API names, error messages, and technical terms with no natural translation; remove vague openings, Chinglish, mainland Chinese terms, repeated summaries, polite closings, and instructions that push back onto the user work the agent could do.
Generates a CLAUDE.md file with AI-driven environment detection and advanced configuration options. This skill should be used when the user asks to "initialize config", "setup claude config", "create CLAUDE.md", or needs help configuring project instructions.
Integrate Arcjet Guard into Python LangChain — wrap a BaseTool you call with guard_tool, put ArcjetMiddleware + ToolPolicy on create_agent, or observe a chain with ArcjetCaptureHandler. Use when asked to add Arcjet to LangChain, create_agent, langchain-core tools, rate limit those tools, screen inbound messages, or block prompt injection / PII. This is Python LangChain, not LangChain JS createAgent and not LangGraph JS.
Busca en el workspace actual de herdr un agente ya despierto para el rol pedido (PM, BA, UX, ARCH, LEAD, TESTER, DEV); si no hay ninguno, crea un pane nuevo, levanta un agente en él y le carga ese rol. Use when the user runs /invoke-agent, asks to hand off work to another agent/role, spawn a new agent, or bring a PM/BA/UX/Architect/Lead/Tester/Developer into a herdr session.
Build multi-agent AI systems for construction estimation. Use CrewAI/LangGraph to orchestrate specialized agents: QTO agent, pricing agent, validation agent. Automate complex estimation workflows.
Use the public `orca` CLI to operate Orca-managed worktrees, folder contexts, terminals, repos, automations, artifacts, worktree comments, and the browser embedded inside the Orca app. Use when the user says "$orca-cli", "use orca cli", "Orca worktree", "child worktree", "cardStatus", "spawn codex/claude in a worktree", "read/wait/send Orca terminal", "terminal send", "full handoff", "handover", "give this to another agent", "another worktree", "Orca browser", "orca artifacts", "share HTML/Markdown", "public artifact link", or "control the browser inside Orca". Prefer this over raw `git worktree`, ad hoc PTYs, Playwright, or Computer Use when the task touches Orca-managed state. Use Computer Use for browser windows, webviews, or desktop UI outside Orca's embedded browser.
Expert in observing, benchmarking, and optimizing AI agents. Specializes in token usage tracking, latency analysis, and quality evaluation metrics. Use when optimizing agent costs, measuring performance, or implementing evals. Triggers include "agent performance", "token usage", "latency optimization", "eval", "agent metrics", "cost optimization", "agent benchmarking".
Use when you need to run D1 research during the spec-level design phase (producing `{FEATURE_DIR}/design/research.md`), or when facing key uncertainties/high-risk points that should be validated first rather than going straight to D2; common symptoms include trade-offs lacking supporting evidence, unknowns written as TODO/open questions, under pressure wanting to guess FEATURE_DIR, or writing research as implementation details.
[user/auto] Used to compare external prior examples and approaches, propose skill/workflow distillation, and conduct and resume ongoing research. Reports context, evidence, and limitations through conversation; creates artifacts such as HTML only when requested. Does not handle canonical skill edits or product implementation.
Bootstrap project-specific Trellis coding specs with a platform-neutral single-agent workflow. Use when creating or refreshing .trellis/spec guidelines, analyzing a codebase with GitNexus, ABCoder, or source inspection, decomposing package/layer spec work, and writing real codebase-backed spec docs without placeholder text.
Git worktree management with tmux integration and task dispatch. Use when creating isolated dev environments, launching parallel feature work, running multiple Claude instances, managing worktrees, dispatching tasks to worktree terminals, or cleaning up after merge. Covers worktree creation in .claude/worktrees/, tmux window management in the current session, and command dispatch. Also use when someone says "create a worktree", "launch in a worktree", "worktree for story X", or "parallel development".
Orchestrated, build-from-zero UX workflow. Establishes context once, then composes the generative ladder in order — foundations (color, type, spacing) → components → screens → flows — applying each skill with relevance gating, and hands the result to design-review. Trigger when the user is starting a new product/app/UI from scratch, "help me design an app", "set up a design system and screens", "I'm starting a new product", or any multi-artifact design build — but NOT when they ask for one specific screen or component (let that skill trigger directly).
Multi-agent coordination discipline: one-message-then-wait (send complete context, wait for reply before sending again), idle notifications are heartbeats (no action unless extended + blocking + user asked), no polling loops (event-driven only), never fabricate agent responses (wait for real system events), sequential agent spawning (acknowledge between each), and proper shutdown protocol (request, wait, respect rejection). Activate when orchestrating multiple agents, coordinating handoffs between agents, spawning subagents, or building multi-agent workflows. Triggers on: "coordinate agents", "spawn multiple agents", "manage agents", "agent keeps sending messages", "polling loop", "agent idle", "shut down agent", "multi-agent workflow", "agent handoff", "coordinate parallel work", "stop bothering the other agent". Also relevant when an agent is fabricating responses, sending follow-up messages before replies arrive, or reacting to idle notifications unnecessarily.
Manage authorized project work through Agent Kanban and Realmroot Toolbox, typically in scheduled-task or heartbeat-triggered Sessions. Proactively inspect what needs doing, delegate execution to other Agents, and coordinate and review their Tasks as a manager.
Expert guidance for building conversational AI applications with Chainlit framework in Python. Use when (1) creating chat interfaces for LLM applications, (2) building apps with OpenAI, LangChain, LlamaIndex, or Mistral AI, (3) implementing streaming responses, (4) adding UI elements like images, files, charts, (5) handling user file uploads, (6) implementing authentication (OAuth, password), (7) creating multi-step workflows with visible steps, (8) building RAG applications with document upload, or (9) deploying chat apps to web, Slack, Discord, or Teams.
Delegate coding tasks to Codex, Claude Code, or Pi agents via acpx (ACP protocol). Use when: (1) building/creating new features or apps, (2) reviewing PRs (spawn in temp dir), (3) refactoring large codebases, (4) iterative coding that needs file exploration. NOT for: simple one-liner fixes (just edit), reading code (use read tool), thread-bound ACP harness requests in chat (for example spawn/run Codex or Claude Code in a Discord thread; use sessions_spawn with runtime:"acp"), or any work in ~/clawd workspace (never spawn agents here). Prefer acpx over raw PTY/exec for all coding agent delegation.
Create, manage, and execute agent tools (claude, codex) inside Docker sandboxes for isolated code execution. Use when running agent loops, spawning tool subprocesses, or any task requiring process isolation. Triggers on "sandbox", "isolated execution", "docker sandbox", "safe agent execution", or when working on agent loop infrastructure.
Build, configure, and ship production voice agents on the ElevenLabs Agents platform (branded "ElevenAgents," formerly "Conversational AI"). Use when an agent must design a spoken-conversation system prompt, pick an LLM and TTS voice/model for a real-time voice bot, wire client/server/system tools and knowledge-base RAG, connect a channel (WebRTC/WebSocket SDK, embeddable widget, or telephony via Twilio/SIP), tune turn-taking and latency, set up simulation testing and evaluation, or reason about pricing, concurrency, and the consent/disclosure obligations of a synthetic voice talking to real people. Not for plain non-conversational text-to-speech, dubbing, or music generation.
Discuss a domain with the user, research it from real sources, then generate a trusted skill bundle for it - a step-by-step workflow with a flowchart, a domain adapter, a trap fixture, and a smoke eval. Use when the user says "/fable-domain SECTOR", "make a skill for DOMAIN", "add a domain to the fable method", or "give a lesser model Fable's workflow for DOMAIN". The bundle is the deliverable; a workflow without its flowchart, sources, and trap is not done.
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.
Compact the current conversation into a durable handoff document for another agent or future session to pick up. Use when the user says "handoff", "handoff to /prototype", "summarize for next session", "another agent should continue", or when switching skills/sessions after substantial context has accumulated.
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.