tranhieutt

Provides vector database and semantic search patterns for Pinecone, Weaviate, Qdrant, Milvus, and pgvector in RAG and recommendation systems. Use when implementing vector search or when the user mentions vector database, semantic search, embeddings, or similarity search.

Orchestrates the backend team of technical-director, backend-developer, data-engineer, and security-engineer to design, implement, and review a backend system end-to-end. Use when a backend feature needs coordinated multi-specialist delivery.

Provides architectural patterns for LLM-powered applications and AI assistants, including prompt engineering, RAG, agent loops, conversation management, and evaluation. Use when building AI-based features, chatbots, or complex AI system architectures.

Forces the strict Red-Green-Refactor development cycle. Requires one failing behavior test through a public interface before each implementation slice.

Orquestra a equipe de lançamento (gerente de lançamento, engenheiro de QA, engenheiro de DevOps e produtor) para executar um lançamento desde a versão candidata até a implantação em produção. Utilize ao coordenar um lançamento entre QA, DevOps e gerenciamento de lançamento.

Gera ou atualiza arquivos de localização i18n, chaves de tradução e configurações de localidade. Utilize ao adicionar um novo idioma, atualizar strings de tradução ou quando o usuário mencionar localização, i18n, tradução ou l10n.

Builds dark-themed TypeScript UIs with accessible color systems, contrast compliance, and responsive design patterns. Use when implementing dark mode or building accessible TypeScript UI components.

Orchestrates the frontend team of ux-designer, ux-researcher, frontend-developer, and accessibility-specialist to design, implement, and validate a UI feature from research to launch. Use when a UI feature needs coordinated design and engineering delivery.

Writes blameless postmortems with root cause analysis, incident timelines, contributing factors, and action items. Use when conducting incident reviews or when the user mentions postmortem, root cause analysis, or blameless review.

Consolidates and prunes the memory directory through a 4-phase reflective pass: Orient, Gather, Consolidate, Prune. Run after intensive sessions, when memories feel duplicated, or when MEMORY.md approaches the 200-line limit.

Strictly enforce context engineering principles to avoid context stuffing, optimize memory architecture, and manage the Research-Plan-Reset-Implement cycle.

Identifies performance bottlenecks, measures against budgets, and generates prioritized optimization recommendations. Use when investigating slow performance or when the user mentions performance profiling, bottleneck, or optimization.