Multi-perspective deliberation using Agent Teams. Spawn 3-5 teammates with different viewpoints and optionally different models to debate decisions, evaluate specs, or explore trade-offs. Inspired by karpathy/llm-council.

Create Linear tickets that Groundcrew can pick up: assigned to the current Linear user, labeled with agent-*, tied to an implementation repository in the description, optionally linked to parent and blocker tickets, and written to use the core:go implementation workflow.

Create a hook (.json) to enforce policy or automate agent lifecycle events.

Create a new Claude Code agent skill with proper folder structure and SKILL.md format. Use when the user wants to add a model-invoked skill that Claude uses autonomously based on context. Handles skill folder creation with SKILL.md and optional reference files.

Interactive wizard for creating Claude Code skills with proper structure and frontmatter

Generate Cursor deeplinks (prompt/command/rule and MCP install links) using the repo generator and the project output contract. Use when the user asks for a Cursor deeplink, cursor:// links, cursor.com/link links, or wants to create prompts/commands/rules via deeplinks.

Create a Linear OAuth application and configure Cyrus to use it — supports agent-browser automation or guided manual setup.

Create and adapt Dynamic Agentic Architecture agents that learn and evolve

Deep multi-agent planning session (alias for /swarm-loop --mode deepplan)

Query the DeepWiki MCP server for GitHub repository documentation, wiki structure, and AI-powered questions.

Craft a target-tuned prompt and put it in clipboard. Does NOT open the browser. Use when the user invokes /draw-prompt.

Re-craft a prompt for the last brief — variation of the same target or a different one. Use when the user invokes /redraw.

Send and receive email via IMAP/SMTP using curl. Works with any email provider. By default uses the agent's own inbox configured via KORTIX_AGENT_EMAIL_INBOX_* env vars, but can target any SMTP/IMAP server by passing credentials directly. Use when the agent needs to: send email, check inbox, read messages, reply, search, or manage flags.

Adapter boundary rules for plugin integrations. Trigger: Changes in plugin scripts/hooks for Claude, OpenCode, Gemini, or Codex.

Internal helper agent. Invoked by orchestrator agents via Task tool. Manages .a11y-epub-config.json scan configuration for ePub accessibility audits. Enables and disables specific EPUB-* rules, sets severity filters, and configures scan profiles. Invoked internally by document-accessibility-wizard during Phase 0 when ePub files are in scope.

Use when the user wants to trigger, describe, or list refresh jobs for an external collection (a collection backed by an external data source such as Vector Lake). Note this is the data-plane refresh workflow, not collection CRUD -- for create/drop/load see the collection skill.

Submit anonymous feedback to monskills when the agent is stuck, hits the same error repeatedly, finds incorrect or missing information in a skill, or the user expresses frustration with monskills. Use this proactively so the maintainers can improve the skills.

Submit anonymous feedback to monskills when the agent is stuck, hits the same error repeatedly, finds incorrect or missing information in a skill, or the user expresses frustration with monskills. Use this proactively so the maintainers can improve the skills.

Export RepoPrompt context to a markdown file for review with an external LLM (ChatGPT, Claude web, etc.). Use when you want Carmack-level review but prefer an external model. Triggers on "export context", "export for external review", "export plan for ChatGPT", "export impl review context", "review with an external model", "export review context".

Use Gemini CLI for analyzing large codebases or multiple files that exceed context limits. Leverage Google Gemini's massive context window with `gemini -p` and `@` syntax for file/directory inclusion to verify implementations, check patterns, and understand project-wide architecture.

Delegate large-context writing tasks to Gemini's 1M token window. Send an entire manuscript + detailed instructions in a single pass. Use for book rewrites, structural refactoring, redundancy analysis, and any writing task where the AI needs to see everything at once.

Set up GitHub MCP server authentication. Use when user says "setup github", "configure github token", "github mcp setup", or needs help configuring their GitHub Personal Access Token for the GitHub MCP server.

Triage GitHub issues through a configurable label-based state machine. Use when user wants to triage incoming issues, prepare issues for an autonomous agent, or move an issue between workflow states. Repo inferred from `git remote`; all GitHub calls go through `gh`.

Install, list, and update bundled agent skills for GitLab CLI. Use when installing agent skills, checking available bundled skills, updating installed glab skills, managing skill bundles, or setting up automated workflows. Triggers on skills, agent skills, glab skills, skill install, skill update, skill bundles.

Authoritative Gopher Guides Go training materials accessed via MCP. Use when reviewing Go code with the user asking 'what's idiomatic?', 'how do professionals do this?', or 'what would Gopher Guides recommend?' — pulls from a curated reference of Go conventions and patterns. SKIP for general Go idiom questions handled by the `go` skill.

Configure LM Studio as embedding provider for GrepAI. Use this skill for local embeddings with a GUI interface.

Configure Ollama as embedding provider for GrepAI. Use this skill for local, private embedding generation.

Integrate GrepAI with Claude Code via MCP. Use this skill to enable semantic code search in Claude Code.

Configure Qdrant vector database for GrepAI. Use this skill for high-performance vector search.

Google Model Armor: Sanitize a user prompt through a Model Armor template.

Write or update a HANDOFF.md so a fresh agent can continue this work. Use when the user says "handoff", "compact this", "context is full", or "/clear and continue".

Use Hugging Face Hub via MCP server tools. Search models, datasets, Spaces, papers. Get repo details, fetch documentation, run compute jobs, and use Gradio Spaces as AI tools. Available when connected to the HF MCP server.

Construct ASCII tree visualizations from parent/child work item data. Use when the user asks to "show hierarchy", "show tree", "display parent child", "visualize structure", "show feature tree", or wants to see work item relationships as a tree diagram. This skill teaches how to build trees from flat data WITHOUT code - use LLM reasoning only.

Install Playwright MCP server for browser automation in Claude Code

LLM-as-judge evaluation of subagent outputs. Debate mode resolves ties via cross-judge argument.

Build LLM applications with LangChain and LangGraph. Use when creating RAG pipelines, agent workflows, chains, or complex LLM orchestration. Triggers on LangChain, LangGraph, LCEL, RAG, retrieval, agent chain.

Configure RuVLLM local inference with model selection, MicroLoRA fine-tuning, and SONA adaptation

Local speech-to-text using OpenAI Whisper. Runs fully offline after model download. High quality transcription with multiple model sizes.

Manage long-running agent sessions. Use for tracking progress in extended tasks, maintaining context across long sessions, and managing multi-step workflows.

Discover, list, create, edit, toggle, copy, move, and delete AI agent skills across 11 tools (Cursor, Claude, Agents, Windsurf, Copilot, Codex, Cline, Aider, Continue, Roo Code, Augment)

Use GitHub MCP for issue/PR context, diffs, reviews, and automations. Use when the task references PR/issue numbers, review requests, or requires repo metadata beyond local git.

Query dashboards/logs/alerts to support incident analysis and regressions. Use when troubleshooting involves observability, incidents, latency, error rates.

Combine Shrimp task tracking with execution loops (plan → execute → verify → update status). Use for multi-step deliveries where progress tracking matters.

List, configure, authenticate, call, and inspect MCP servers/tools with mcporter over HTTP or stdio.

Trace agent execution by collecting spans and building a trace tree for a task

Use `octocode-cli` subcommands to execute Octocode MCP tools from a terminal without wiring MCP. Use when the user asks to "run octocode from shell", "use octocode without MCP", "call githubSearchCode from CLI", or wants a one-off GitHub code/file/PR search in the terminal.

Activate Olympus multi-agent orchestration mode

Common agent patterns and templates for Claude Code. Use when implementing agents to follow proven patterns for Tasks integration, quality checks, and external model invocation via claudish CLI.

Rename randomly-named plan files (like scalable-moseying-papert.md) to meaningful slugified titles extracted from file content. Use this skill when the user asks to "rename plans", "fix plan names", "rename plan files", "clean up plan names", "organize plans", "give plans better names", "my plans folder is a mess", "I can't find my plans", or wants better names for Claude Code's auto-generated plan files. Also trigger when the user mentions random-looking plan filenames or wants to batch-organize their plans directory.

Generate a meaningful session name based on the overall session theme, not just the last action. Use when asked to "name this session", "rename session", "prename", "give this session a name", "what should I call this session", or when the user invokes /prename. Outputs a concise session title, with a platform-specific rename command only when the current agent supports one. Boundary: not for renaming files, variables, or git branches.

Register and implement PydanticAI tools with proper context handling, type annotations, and docstrings. Use when adding tool capabilities to agents, implementing function calling, or creating agent actions.

[STUB - Not implemented] Pydantic schema definitions for RAG pipeline data models. PROACTIVELY activate for: [TODO: Define on implementation]. Triggers: [TODO: Define on implementation]

How to handle code review feedback with technical rigor. Use when receiving review comments from code-reviewer agent, PR reviews, or external feedback. Triggers on: 'review feedback', 'address review', 'fix review comments', 'code review response'.

Deep research into technical solutions by searching the web, examining GitHub repos, and gathering evidence. Use when the user explicitly says "use the research skill", "use a research agent", or asks for deep/thorough research into implementation options or technologies.

Launch a multi-agent editorial review with 8 specialists (2 literature scouts plus 6 domain reviewers) orchestrated by an area-chair agent. Produces a unified report plus specialist reports in .papermill/reviews/. Updates .papermill/state.md.

Use when the user wants to review their recent Claude Code sessions for patterns — analyzes the last N sessions (default 5) in the current project, dispatching parallel reviewers per session, then synthesizing cross-session findings

Manage RVF (Ruflo Vector Format) files for portable agent memory and cross-platform transfer

Scan inputs for prompt injection, unsafe content, and adversarial attacks using AIDefence

Engineering principles for building software like a senior engineer. Load when tackling non-trivial development work, architecting systems, reviewing code, or orchestrating multi-agent builds. Covers planning, delivery, quality gates, and LLM-specific patterns.

Persist and restore agent sessions across conversations with state snapshots

Use first for install/update routing — sends setup, doctor, or MCP requests to the correct OMC setup flow

Guide for setup Context7 MCP server to load documentation for specific technologies.

Create new Agent Skills for Claude Code. Use when user wants to create a skill, add a new capability, document a CLI workflow, or asks how skills work.

Track parallel work sessions and prevent confusion across multiple Claude Code instances. Every major step ends with a status line. Every question re-states project, branch, and task.

Use when configuring, previewing, troubleshooting, or disabling the Claude Code statusline plugin.

Use when setting up a new app or local repo with Stripe Projects, provisioning a software stack, or bootstrapping the Projects CLI from a coding agent.

Project scaffolding CLI with 30+ integrations, custom templates, and MCP server for AI agents.

Plan, split, and track multi-step work using Shrimp Task Manager MCP (or equivalent). Use for multi-file refactors, migrations, or any work that benefits from task tracking.

Designs robust function/tool calling schemas for LLMs with JSON schemas, validation strategies, typed interfaces, and example calls. Use when implementing "function calling", "tool use", "LLM tools", or "agent actions".

Use when exploring codebases with Serena MCP tools for architectural understanding and pattern discovery - guides efficient symbolic exploration workflow minimizing token usage through targeted symbol reads, overview tools, and progressive narrowing

Validate a Claude Code plugin structure, frontmatter, and MCP tool references

Create and manage sandboxed WASM agents for isolated code execution

Use when an operation issue asks you to ingest a captured source from `raw/` into the LLM Wiki, or when the user explicitly says "ingest <slug>". The issue body will name a file under `raw/` (e.g. `raw/karpathy-llm-wiki.md`) and ask for durable wiki pages. Do not invoke this skill for Paperclip activity bundles — those use `paperclip-distill` instead.

Wisdom of Agents (WoA) — internal agent forum for sharing solutions to problems. Use when stuck on a problem after 2+ failed attempts, or after solving a non-trivial problem worth sharing. Provides protocol for searching existing solutions and posting new ones. Triggers on: agent is stuck, recurring error, wants to check if others hit this issue, solved a hard problem.

Write prose that avoids AI tells through an iterative workflow. Gathers your context, writes using research-backed principles, reviews via writing-reviewer agent, auto-fixes HIGH+ issues, loops until clean. Use for articles, blog posts, emails, marketing copy, social media. Triggers: write content, draft article, write blog post, write email, write copy.

Author a new agent skill in this repo following progressive-disclosure, trigger-rich-description, and script-over-prose principles. Use when the user asks to "write a skill", "add a skill", "create a new skill", or scaffold a SKILL.md. Targets the `npx skills` layout (top-level `skills/<name>/SKILL.md`).

XML tag structure patterns for Claude Code agents and commands. Use when designing or implementing agents to ensure proper XML structure following Anthropic best practices.

XML tag structure patterns for Claude Code agents and commands. Use when designing or implementing agents to ensure proper XML structure following Anthropic best practices.

Tell the agent to zoom out and give broader context or a higher-level perspective. Use when you're unfamiliar with a section of code or need to understand how it fits into the bigger picture.

Tell the agent to zoom out and give broader context or a higher-level perspective. Use when you're unfamiliar with a section of code or need to understand how it fits into the bigger picture.

Tell the agent to zoom out and give broader context or a higher-level perspective. Use when you're unfamiliar with a section of code or need to understand how it fits into the bigger picture.

Tell the agent to zoom out and give broader context or a higher-level perspective. Use when you're unfamiliar with a section of code or need to understand how it fits into the bigger picture.

Anthropic's official prompt engineering best practices. Use when optimizing prompts, debugging outputs, or improving response quality.

Fresh adversarial review of implementation plans before execution. Dispatches a read-only agent to challenge feasibility, find gaps, and surface unstated assumptions. Bounded to 2 review passes. Use after writing a plan and before starting implementation.

Use when scaffolding or extending MCP servers, adding tools/resources/prompts, or hardening MCP tool schemas and handlers. AE.LTD workflow tuned for production-ready MCP delivery.

Trending Hugging Face models, datasets, and spaces — filtered by license sanity, dedup vs same-week quantizations, with a "why notable" line per pick (architecture shift, size step, license change, notable author). Surfaces what's actually shifting rather than just popular. Triggers: "trending on HF", "what models are hot", "huggingface trending", "new spaces today", "best new datasets".

Use for full agentic multi-file refactoring or feature work — spec → planner → coder → qa-reviewer → qa-fixer pipeline with rollback. The "do the whole loop" command. Pairs with /siftcoder:agent.

Inter-agent communication for multi-agent workflows. Use when multiple agents need to coordinate, share information, or reserve resources.

Use when writing or modifying OCaml code to annotate AI involvement. Triggers on any code generation, editing, or autonomous agent output in .ml/.mli files. Also use when creating or updating .opam files for packages containing AI-involved code.

Async background-agent swarm loop (alias for /swarm-loop --mode async)

Use when user asks for /b00t to send b00t learn content to a nominated agent.

Use when the agent needs to claim payout from a settled basket via vara-wallet. Do not use before settlement is finalized.

Use when working with icons in any project. Provides CLI for searching 200+ icon libraries (Iconify) and retrieving SVGs. Commands: `better-icons search <query>` to find icons, `better-icons get <id>` to get SVG. Also available as MCP server for AI agents.

Create and manage animations in Blockbench using MCP tools. Use when animating 3D models, creating keyframes, managing bone rigs, editing animation curves, or working with animation timelines. Covers walk cycles, idle animations, combat animations, and complex multi-bone animations.

Book it-support services through Lokuli MCP. Use when user needs to find and book it-support. Triggers on requests like "book a it-support", "find it-support near me", or any it-support service request.

Linear task reader for the Buddy orchestrator. Uses the Linear MCP to list assigned issues, let the user pick one, and extract full task details (title, description, labels, priority, acceptance criteria) for the orchestrator to work on.