Desenvolvimento

Sequential subagent execution with two-stage review gates for implementation plans. Use when executing multi-task plans in current session, when tasks need fresh subagent context to avoid pollution, when formal review cycles (spec compliance then code quality) are required between tasks, or when you need diff-based validation of each task before proceeding.

When the user needs to plan a complex feature, system, or project that requires deep creative exploration before execution. Use when the user says "plan this," "let's design," "how should we build," "architect this," "brainstorm then plan," "ultraplan," "deep plan," "think this through," or when a task is too complex for a simple plan. Combines creative brainstorming with multi-agent technical val

Save the current in-conversation version of loaded skills as a new plugin. Use when the user says "save these skills", "save my iterated skills", "save loaded", "package what I've been working on", "save this as a plugin", "install what I loaded", "I forgot to click save plugin", or runs /skills-save. Captures the CURRENT state from this conversation — iterated versions if any, originals otherwise

Use when writing libGDX Java/Kotlin code involving collections (Array, ObjectMap, ObjectSet, Queue), object pooling (Pool, DefaultPool), JSON serialization (Json, JsonReader, JsonValue), I18NBundle localization, Timer scheduling, or utility classes (Align, Scaling, TimeUtils, AsyncExecutor). Use when debugging GC pressure, missing identity parameter, wrong map iteration, pooling leaks, JSON parse

Store policy compliance checker for iOS App Store and Google Play. Validates metadata against Apple App Store Review Guidelines and Google Play Developer Program Policies. Catches violations that would cause rejection. Also validates ASO recommendations from other agents. Triggers on: "compliance", "policy check", "guidelines", "store policies", "review guidelines", "will this get rejected", "is t

When the user wants to deeply understand a codebase, map its architecture, or build a persistent mental model for future sessions. Use when the user says "explain this codebase," "map the architecture," "codebase insight," "understand this repo," "what does this project do," "how does this work," "analyze the codebase," or when starting work on an unfamiliar project. Produces a persistent insight

Explains code with visual diagrams and analogies. Use when explaining how code works, teaching about a codebase, or when the user asks "how does this work?"

Use when facing a bug, test failure, or unexpected behavior that isn't immediately obvious

Avalia LLMs em mais de 60 benchmarks acadêmicos como MMLU e HumanEval. É um padrão da indústria usado por EleutherAI, HuggingFace e grandes laboratórios para comparar a qualidade de modelos, relatar resultados acadêmicos e acompanhar o progresso de treinamento, com suporte para HuggingFace, vLLM e APIs.

Use when writing libGDX Java/Kotlin code involving AssetManager — loading assets (Texture, TextureAtlas, Sound, Music, BitmapFont, Skin, Model, TiledMap, ParticleEffect, ShaderProgram, I18NBundle), async loading screens, reference counting, screen transitions, custom loaders, or FreeType font loading via AssetManager. Use when debugging assets not loading, double-dispose crashes, missing textures

Write skillprobe YAML tests for LLM skills. Use when asked to write tests, create test scenarios, test a skill, generate skillprobe tests, or check whether a skill activates correctly.

Canonical overlay for non-trivial server-side networked code that needs multi-layer architecture, data-access, reliability, trust-boundary, or deeper testing discipline. Use alongside the repo's implementation skill for backend services, APIs, middleware, queues, repositories, transactions, auth-sensitive flows, or backend refactors; use `backend-guidance` for thin routine handler work.