Desenvolvimento

17 habilidades do Claude Code para contexto de IA, testes e desenvolvimento de MCP. Utiliza um formato registrado pela IANA (application/vnd.faf+yaml) para criar o DNA do projeto, pontuar a prontidão da IA, sincronizar com CLAUDE.md, construir servidores MCP e gerar suítes de teste, com mais de 36.000 downloads em npm, PyPI e crates.io.

Workflow for drafting, rewriting, splitting, and auditing story-level requirements, user stories, feature definitions, definitions of done, and acceptance criteria into unambiguous story cards for humans and coding agents. Use for rough tickets, broad epics, or unclear product behavior; avoid for full PRDs, glossary definitions, and code implementation.

Use when writing libGDX Java/Kotlin code involving FrameBuffer (FBO), render-to-texture, post-processing, FloatFrameBuffer, GLFrameBuffer, Pixmap, runtime texture generation, screen capture, pixel manipulation, or PixmapIO. Use when debugging upside-down FBO rendering, FBO nesting issues, native memory leaks from Pixmap, or blend state corruption from post-processing.

Dispatches one subagent per independent domain to parallelize investigation/fixes. Use when you have 2+ unrelated failures (e.g., separate failing test files, subsystems, bugs) with no shared state or ordering dependencies.

Activate or deactivate Rocky's Eridian conversation style — clipped grammar, no articles, tripled emphasis, engineering vocabulary. Use when the user says /rocky-talk, "turn on Rocky talk", "speak like Rocky", or "turn off Rocky talk". Changes how the agent communicates, not how it thinks.

Register the domain's feature flag in `feature_flag.xml`. This entry populates the `ps_feature_flag` table at install/upgrade and enables the flag-based routing between legacy and Symfony controllers.

Use when writing libGDX Java/Kotlin code involving 2D or 3D particle effects — ParticleEffect, ParticleEffectPool, ParticleEmitter, particle loading, AssetManager integration, blend functions, or the 3D particle system (ParticleSystem, BillboardParticleBatch). Use when debugging frozen particles, compounding scale, pooled effect leaks, or wrong blend modes.

TDD-first spec management for AI coding workflows, producing canonical SPEC.html files with red-green-refactor state. Use this companion when the user explicitly mentions specs, forging, structured planning, resume/status, implementing a spec, red-green-refactor, or running tests. Do NOT trigger on ordinary coding tasks that do not ask for spec management.

Use when the user says "init memory", "set up memory", "scaffold memory", "init project memory", or invokes /init-project-memory — creates the .claude/ session memory system in a new project

Activate or deactivate full Rocky mode (talk style + engineering mind together). Use when the user says /rocky, "turn on Rocky", "activate Rocky", or "turn off Rocky".

Disciplined plan execution for implementation tasks. Use when executing a saved implementation plan, following step-by-step instructions from a plan document.

Iteratively enriches incomplete JIRA Epics into estimation-ready, technically-validated specs. Five modes: batch readiness scan, RD discovery (Phase 0), PM elaboration (Phase 1), technical approach (Phase 2), and multi-round iteration. Phase 1 goes beyond checklist filling — it explores the codebase, hardens AC, and produces a structured artifact for downstream skills. Trigger: "refinement", "groo