Ruta de Aprendizaje
s01 a s12: Diseño Progresivo del Agente
Leyenda de Capas
The Agent LoopBash is All You Need
“The minimal agent kernel is a while loop + one tool”
Saber MásToolsOne Handler Per Tool
“The loop stays the same; new tools register into the dispatch map”
Saber MásTodoWritePlan Before You Act
“An agent without a plan drifts; list the steps first, then execute”
Saber MásSubagentsClean Context Per Subtask
“Subagents use independent messages[], keeping the main conversation clean”
Saber MásSkillsLoad on Demand
“Inject knowledge via tool_result when needed, not upfront in the system prompt”
Saber MásCompactThree-Layer Compression
“Context will fill up; three-layer compression strategy enables infinite sessions”
Saber MásTasksTask Graph + Dependencies
“A file-based task graph with ordering, parallelism, and dependencies -- the coordination backbone for multi-agent work”
Saber MásBackground TasksBackground Threads + Notifications
“Run slow operations in the background; the agent keeps thinking ahead”
Saber MásAgent TeamsTeammates + Mailboxes
“When one agent can't finish, delegate to persistent teammates via async mailboxes”
Saber MásTeam ProtocolsShared Communication Rules
“One request-response pattern drives all team negotiation”
Saber MásAutonomous AgentsScan Board, Claim Tasks
“Teammates scan the board and claim tasks themselves; no need for the lead to assign each one”
Saber MásWorktree + Task IsolationIsolate by Directory
“Each works in its own directory; tasks manage goals, worktrees manage directories, bound by ID”
Saber Más