Các tầng kiến trúc
Năm mối quan tâm độc lập ghép thành một agent hoàn chỉnh
L1 Tools & Execution
Agent CÓ THỂ làm gì. Nền tảng: tool cho mô hình khả năng tương tác với thế giới.
The Agent Loop
One Loop Is All You Need
The smallest useful agent is a loop that calls the model, runs tools, and feeds results back.
Tool Use
Add a Tool, Add Just One Line
The loop stays stable while capabilities register into a dispatch table.
Permission
Check Permissions Before Execution
Dangerous actions need a harness decision point before the shell runs.
Hooks
Hang on the Loop, Don't Write into It
Cross-cutting behavior belongs around the loop, not tangled inside it.
L2 Planning & Control
Cách tổ chức công việc. Từ danh sách việc đơn giản đến bảng task có phụ thuộc, dùng chung giữa các agent.
TodoWrite
An Agent Without a Plan Drifts Off Course
Explicit plans keep long-running work visible and correctable.
Subagent
Break Large Tasks into Small Ones with Clean Context
Subagents give each subtask a clean message history while preserving the main thread.
Skill Loading
Load Only When Needed
Inject specialized knowledge only when the task actually needs it.
Goal Loop
Independent Evaluation Decides When to Stop
A durable goal keeps the loop working until an independent evaluator finds the completion condition satisfied in the conversation.
L3 Memory Management
Giữ ngữ cảnh trong giới hạn. Chiến lược nén giúp agent làm việc lâu mà không mất mạch.
L4 Concurrency & Scheduling
Thực thi không chặn. Luồng nền và bus thông báo cho công việc song song.
Background Tasks
Slow Operations Go to the Background
The agent can keep reasoning while slow work completes elsewhere.
Cron Scheduler
Producing Work on a Schedule
Recurring work should be created by the harness, not remembered by the model.
Workflow Runtime
Scripts Own Fixed Orchestration
When orchestration has a fixed shape, code can make it parallel, deterministic, and resumable.
L5 Multi-Agent Platform
Phối hợp đa agent. Đội, nhắn tin, nhận task nguyên tử, và thư mục worktree gắn với task.
Task System
Break Big Goals into Small Tasks
A task graph turns vague goals into ordered, observable work.
Agent Team Runtime
Persistent Teammates, Atomic Claims, Task-Bound Worktrees
Persistent teammates can reliably discover and execute parallel work when the runtime owns messaging, atomic claims, and task-bound working directories.
MCP Tools
External Tools, Standard Protocol
External services can become agent tools through a standard discovery and call protocol.
Integrated Harness
Many Mechanisms, One Loop
The integrated harness is still one loop, surrounded by the systems introduced across the course.