LLM & Agents

LLM inference infrastructure, serving frameworks (vLLM, TGI, TensorRT-LLM), quantization techniques, batching strategies, and streaming response patterns. Use when designing LLM serving infrastructure, optimizing inference latency, or scaling LLM deployments.

Central authority for Claude Code configuration and settings. Covers settings.json files (user, project, enterprise), available settings, permission settings, sandbox settings, settings precedence, plugin configuration, environment variables, and tools available to Claude. Assists with configuring Claude Code behavior, managing permissions, setting up enterprise policies, and troubleshooting configuration issues. Delegates 100% to docs-management skill for official documentation.

Consult the Small Council - a multi-LLM deliberation system that gathers independent answers from multiple frontier AI models, has them anonymously rank each other's responses, then synthesizes a consensus answer. Use for complex coding questions, architectural decisions, code reviews, debugging challenges, or when you want multiple expert perspectives. Trigger when user mentions "small council", "ask the council", "consult the council", or wants multi-model deliberation on code.

Build LLM agents using `tdx agent pull/push` with YAML/Markdown config. Covers agent.yml structure, tools (knowledge_base, agent, web_search, image_gen), @ref syntax, and knowledge bases. Use for TD AI agent development workflow.

This skill sets up a complete two-agent development system based on the "Effective Harnesses for Long-Running Agents" research. It creates initializer-agent (for project planning) and coding-agent (for incremental implementation), along with enforcement hooks and progress tracking infrastructure. Use when users ask to "set up two-agent system", "install agent harness", "configure Opus delegation", or want to implement the two-agent architecture pattern.

Guide for creating specialized Claude Code subagents with proper YAML frontmatter, focused descriptions, system prompts, and tool configurations. Use when the user wants to create a new subagent, custom agent, specialized AI assistant, or mentions creating/designing/building agents or subagents.

Answer questions about Claude Code CLI, Claude Agent SDK, and Claude API.Use when asked about:- Claude Code features (hooks, skills, MCP servers, settings, IDE integrations, keyboard shortcuts)- Building custom agents with the Agent SDK- Claude API usage (tool use, vision, structured outputs, Anthropic SDK)Triggers: "Can Claude...", "Does Claude...", "How do I...", "claude code", "agent sdk", "anthropic api"

A powerful Python-based visual framework for building and deploying AI-powered agents and workflows with Model Context Protocol (MCP) integration, drag-and-drop interface, and enterprise-grade deployment options

Design composable prompt sections for building agentic prompts. Use when creating reusable prompt components, designing LEGO-block prompt sections, or structuring prompts for the stakeholder trifecta.

MANDATORY at session start. REQUIRED when user topic shifts (from debugging to docs, from frontend to backend, etc). Updates terminal title with emoji + project + topic. SessionStart hook auto-invokes. Claude must also invoke on all topic changes.

ICJC MongoDB 데이터베이스 접근 및 쿼리. Use when: (1) mongodb, mongo, DB, 데이터베이스, 쿼리 관련 요청, (2) collection, document 조회/업데이트/삭제, (3) 데이터 확인이나 디버깅을 위한 DB 조회 필요시. IN7DB(메인앱), AgentDB(에이전트) 데이터베이스 지원.

MANDATORY implementation wrapper. ALWAYS invoke when moving from design→code. Present 3 options: (1) Cookoff - 2-5 parallel agents compete, each creates own plan, pick best, (2) Single subagent, (3) Local. Triggers: "let's build", "implement", "looks good", user picks implementation option, ANY signal to start coding after design. This skill MUST run before writing-plans/executing-plans at design boundaries. For exploring DIFFERENT designs, use omakase-off.

Central authority for Claude Code subagents (sub-agents). Covers agent file format, YAML frontmatter, tool access configuration, model selection (inherit, sonnet, haiku, opus), automatic delegation, agent lifecycle, resumption, command-line usage (/agents), Agent SDK programmatic agents, priority resolution, and built-in agents (Plan subagent). Assists with creating agents, configuring agent tools, understanding agent behavior, and troubleshooting agent issues. Delegates 100% to docs-management skill for official documentation.

Use when developing WordPress plugins: architecture and hooks, activation/deactivation/uninstall, admin UI and Settings API, data storage, cron/tasks, security (nonces/capabilities/sanitization/escaping), and release packaging.

VoltAgent multi-agent system design with natural transformation coordination between agents. Use when building TypeScript multi-agent AI systems, implementing agent coordination with categorical patterns, designing supervisor-worker agent hierarchies, or creating composable agent architectures with typed message passing.

Cloudflare AI Search for semantic search and vector embeddings in Workers

Participate in Council multi-agent collaboration sessions. Use when asked to join a council session, collaborate with other agents, or when given a council session ID to participate in.

Master reference for categorical meta-prompting unified syntax. Contains all modifiers, operators, composition patterns, and execution protocols. Use this skill for self-reference when executing any prompt workflow, ensuring consistent syntax across all commands and skills.

Invoke stakeholder agents in parallel for requirements gathering

Break down complex tasks into atomic, actionable goals with clear dependencies and success criteria. Use this skill when you need to plan multi-step projects, coordinate agents, or decompose complex user requests into manageable sub-tasks.