Learning Path
13 progressive chapters, from bootstrap to production engineering
Agent Loop
Heart of the conversation loop
“Every Agent is essentially a loop: call model → execute tool → return result”
Tool System
Registry & dispatch of 50+ tools
“Register a handler, gain a capability — the loop never changes”
Prompt Engineering
Dynamic assembly pipeline
“System Prompt is not a string — it's a dynamically assembled pipeline”
Shell Security
300KB+ security validation
“The most powerful tool needs the tightest security”
Permission Engine
Every operation is checked
“Permissions are not an afterthought — they are the skeleton of the architecture”
Context Management
Infinite work in a finite window
“Context always fills up — the key is how to compress”
MCP Protocol
Unified tool calling standard
“MCP lets any service become an AI tool”
Plugin Ecosystem
Extensible capability boundary
“Plugins are capability multipliers”
Multi-Agent
Agent/Team/Swarm
“Scale comes from division of labor, not larger context”
CLI Transport
Bridge from terminal to remote
“The transport layer determines where an Agent can run”
Bootstrap
From Enter to prompt
“Fast path determines experience, full path determines capability”
Production Patterns
Demo → Production
“Making an Agent reliable requires 10x engineering effort”
Hidden Features
Feature flag-gated hidden modules
“Behind every feature('FLAG') lies an unreleased product decision”