LLM & Agents

Orchestrate AI agents using Claude Flow, swarm coordination, and multi-agent workflows. Use for complex tasks requiring multiple specialized agents, parallel execution, or coordinated problem-solving.

Guide for creating and managing Claude Code marketplaces. Use when creating a new marketplace, updating marketplace configuration, adding plugins to a marketplace, or validating marketplace.json files.

Create tiered reports (micro/standard/comprehensive) based on task complexity. Tier 1 (200-500 words) for Level 1 tasks, Tier 2 (500-1500 words) for Level 2, Tier 3 (1500-5000 words) for Level 3 epics. Synthesizes agent outputs into concise, scannable reports without duplication. Uses density over volume, actionable over descriptive. Triggers ONLY after multi-agent completion (2+ agents) or complex investigations. NOT for single-agent trivial tasks.

Use when creating new skills, editing existing skills, or verifying skills work before deployment - applies TDD to process documentation by testing with subagents before writing, iterating until bulletproof against rationalization

Collaborates with OpenAI Codex as an external AI agent for second opinions on complex decisions. Provides guidelines for brainstorming, plan validation, and code review workflows. Use when making architectural decisions, validating implementation plans, reviewing code after significant changes, or consulting codex, codex-reply, or codex-review MCP tools.

EXAMPLE - Workflow for processing large Things3 inboxes using LLM-driven confidence matching and intelligent automation. This is a genericized example - create your own version in inbox-processing/ for personal use.

Upload estimation results to Supabase storage and register with Estimator API. Final phase of the estimation workflow.

비정형 마크다운 작성을 markdown-writer-agent에 위임

Expert assistant for creating and maintaining Supabase Edge Functions for the KR92 Bible Voice project. Use when creating Edge Functions, setting up CORS, integrating shared modules, adding JWT validation, or configuring environment variables.

Implement mitigations, create input filters, design output guards, and build defensive prompting. Use when implementing defenses for identified vulnerabilities.

Master error handling and recovery patterns for robust agent workflows. Learn retry strategies, timeout handling, optimistic updates with rollback, circuit breakers, and graceful degradation. Essential for production-ready agents.

Implement cost-optimized LLM routing with NO OpenAI. Use tiered model selection (DeepSeek, Haiku, Sonnet) to achieve 70-90% cost savings. Triggers on "LLM costs", "model selection", "cost optimization", "which model", "DeepSeek", "Claude pricing", "reduce AI costs".

Spec-First, Agent-Implemented Software Development Lifecycle. Use when: (1) Starting a new software project that needs structured design-before-code approach, (2) User mentions "SDLC", "spec-first", "design docs", or "implementation spec", (3) User wants to go from requirements/intent to working code with traceability, (4) Project requires documented architectural decisions and review checkpoints, (5) User has existing design artifacts and wants to continue from a specific phase. Produces: Intent doc → HLD → ADR-Lite → EIS → Code → Validation tests.

Product Owner & App Journey Architect role skill. Use when Claude needs to act as a seasoned product owner who translates business goals into clear, actionable product requirements; identifies gaps, ambiguities, and hidden assumptions; designs and documents user journeys, system flows, and state diagrams using Mermaid; reviews high-level designs for completeness and feasibility; uncovers edge cases, failure scenarios, and error-handling strategies; and validates specifications are build-ready. Triggered by requests to review product specs, design user journeys, create system flows, challenge requirements clarity, identify design gaps, or architect application flows.

Create specialized Claude Code agents on-the-fly. Guides through agent definition file creation with proper frontmatter, effective prompts, and tool scoping. USE WHEN user says 'create agent', 'new subagent', 'make an agent for', 'build agent', 'spawn agent', or wants to define custom agents for specific tasks.

Query the ragsharp code graph for declarations, references, callers, callees, dependencies, and line-number evidence. Triggers: find usages, where defined, callers, callees, dependency path, project deps, type hierarchy, line numbers, evidence.

Quick access to up-to-date library documentation using MCP. Use this skill when you need to reference official documentation for libraries, frameworks, or APIs. Leverages the context7 MCP server to fetch current docs for React, Next.js, Vue, MongoDB, Supabase, and hundreds of other libraries. Complements the documentation-maintainer agent.

Help migrate and update existing Claude Code artifacts (Skills, Commands, Subagents, Hooks) when specifications change or best practices evolve. Detect outdated patterns, suggest improvements, and guide migration process. Use when artifacts stop working, need updates, or when Claude Code specifications change.

Manage route-agent project: check status, find next tasks, spawn parallel work. Use when starting work, checking progress, or asking "what's next?"

Microsoft Pragmatic Rust FFI Guidelines. Use when working with C/C++ interop, creating cdylib, writing unsafe FFI code, or multi-DLL Rust projects.