Agent Engineering
From autocomplete user to superstar developer — coding agents, context, skills, MCP, hooks, spec-driven and parallel workflows, and building your own agents.
For students and working developers who want to go from autocomplete to the developer who multiplies a team. Works with Cursor, Claude Code, Codex, Copilot, and OpenClaw.
10 chapters · 62 lessons · 10 labs · 60 quiz questions
- 1
Find your level
A two-minute habit check places you on the five-level ladder and picks your next three moves.
Take the check →
- 2
Work through the stages
Ten chapters in four stages, one lesson at a time, each ending in a quiz and a lab.
See the path →
- 3
Prove it
Ship a capstone repo with a custom agent, guardrails, and evals — scored on a rubric.
View the capstone →
Learning path
Each stage lines up with a level on the ladder.
- Stage 1FoundationsLevel 1 → 2Pick a tool, learn the agent loop, and give it the right context.
- 01The AI Developer WorkstationCursor, Claude Code, Codex, and Copilot — what each surface is for, how agent loops work, and the daily habits that separate fast from reckless.6 lessons
- 02Context EngineeringAGENTS.md, CLAUDE.md, Cursor rules, and Copilot instructions — what belongs in always-on memory, what belongs in a skill, and how to spend the token budget.6 lessons
- Stage 2Work like a superstarLevel 3Specs, tests first, parallel agents, and reusable skills and subagents.
- 03Superstar PracticesThe habits that separate 10× from busy: spec-driven development, TDD with agents, parallel agents on worktrees, model routing and cost, debugging with evidence, and agents in CI.7 lessons
- 04Agent SkillsThe open SKILL.md standard: progressive disclosure, description-driven triggering, scripts and references, evaluation, and trusting third-party skills.6 lessons
- 05Plugins, Commands & SubagentsChoosing between slash commands, skills, subagents, and plugins — and packaging them so a whole team gets the same setup.6 lessons
- Stage 3Connect & orchestrateLevel 4Plug agents into real systems and run durable, multi-step workflows.
- 06Model Context Protocol (MCP)Tools, resources, and prompts over stdio and Streamable HTTP; the stateless 2026 revision; OAuth; configuring MCP in every client; and designing lean, safe servers.6 lessons
- 07OpenClaw & Always-On AgentsThe self-hosted gateway pattern: channels, sessions, skills precedence, ClawHub, plugin bundles, migration from Claude Code, and delegating coding to CLI agents.6 lessons
- 08LangGraph, LangChain & Agent WorkflowsWhen to build a workflow vs an agent, LangGraph state/reducers/checkpoints, interrupts and human approval, LangChain create_agent middleware, and Deep Agents.6 lessons
- Stage 4Build & shipLevel 5Build your own agents and ship agent work safely to a team.
- 09Build Your Own AgentsGo from using agents to shipping them: Claude Agent SDK, OpenAI Agents SDK, Cursor SDK, browser and computer use, A2A for agent-to-agent, and running agents in production.7 lessons
- 10Hooks, Security, Evals & ShippingDeterministic guardrails with hooks, permission and sandbox design, reviewing agent output, evals for agent setups, CI and cloud agents, and rolling this out to a team.6 lessons
Practice & reference
Hands-on labs
10 labs · 45–120 minBuild real artifacts: context files, skills, hooks, parallel worktrees, an SDK agent.
Concepts & trainer
14 scenariosCompare tools, learn the four layers, and practice picking the right primitive.
Starter kit
22 templatesCopy-paste AGENTS.md, hooks, MCP configs, and CI workflows filtered to your tools.
FAQ
53 answersStraight answers on tools, cost, security, SDKs, and team rollout.
Practice quizzes (60) · mastery syncs with curriculum and CS dashboard