Camadas Arquiteturais
Cinco aspectos ortogonais que compoem um agente completo
L1 Tools & Execution
O que o agente PODE fazer. A base: ferramentas dao ao modelo capacidades para interagir com o mundo.
L2 Planning & Coordination
Como o trabalho e organizado. De listas todo simples a quadro de tarefas com dependencias compartilhadas entre agentes.
TodoWrite
Plan Before You Act
An agent without a plan drifts; list the steps first, then execute
Subagents
Clean Context Per Subtask
Subagents use independent messages[], keeping the main conversation clean
Skills
Load on Demand
Inject knowledge via tool_result when needed, not upfront in the system prompt
Tasks
Task Graph + Dependencies
A file-based task graph with ordering, parallelism, and dependencies -- the coordination backbone for multi-agent work
L3 Memory Management
Mantendo o contexto dentro dos limites. Estrategias de compressao que permitem aos agentes trabalhar infinitamente sem perder coerencia.
L4 Concurrency
Execucao nao bloqueante. Threads em segundo plano e barramentos de notificacao para trabalho paralelo.
L5 Collaboration
Coordenacao multi-agente. Times, mensageria e companheiros autonomos que pensam por conta propria.
Agent Teams
Teammates + Mailboxes
When one agent can't finish, delegate to persistent teammates via async mailboxes
Team Protocols
Shared Communication Rules
One request-response pattern drives all team negotiation
Autonomous Agents
Scan Board, Claim Tasks
Teammates scan the board and claim tasks themselves; no need for the lead to assign each one
Worktree + Task Isolation
Isolate by Directory
Each works in its own directory; tasks manage goals, worktrees manage directories, bound by ID