Module 6: AI Workflow Tools
Duration: 90 minutes Day: Day 2, Session 2
What This Module Is About
You've been using AI tools as a smarter autocomplete. This module changes that. You'll work with structured AI workflows — tools that enforce discipline, track decisions, and produce production-grade artifacts instead of one-off suggestions.
The central skill is context engineering: not writing better questions, but structuring the right context so that AI tools consistently produce expert-level output for your specific infrastructure.
You'll also set up cross-session memory, so the context you build in one session automatically carries into the next — and learn how to choose between quick AI-assisted tasks and full structured workflows.
Learning Objectives
By the end of this module, you will be able to:
- Apply the GSD structured workflow for multi-file IaC tasks — from requirements gathering through execution and verification
- Create and use a CLAUDE.md as system context for your project — the artifact that makes every AI interaction namespace-aware, version-specific, and constraint-respecting
- Configure cross-session memory using claude-mem (Claude Code) or MCP memory server (Crush), so past decisions surface automatically in new sessions
- Select the right plan mode for the task at hand — Claude Code
/planfor quick single tasks, GSDplan-phasefor multi-file production work - Extend AI tools with disciplined workflows — understand how TDD, systematic debugging, and code review become more powerful with AI assistance
Prerequisites
- Modules 1–5 completed
- Claude Code or Crush installed and connected to an LLM
- KIND cluster with Prometheus and Grafana deployed (from Module 5 setup)
- Familiarity with the reference app (api-gateway, catalog, worker services)
Module Contents
| Section | Content | Time |
|---|---|---|
| Lab — Section 1 | GSD Workflow: Build a Monitoring Stack | 30 min |
| Lab — Section 2 | Context Engineering Practical | 20 min |
| Lab — Section 3 | Memory Systems | 15 min |
| Lab — Section 4 | Plan Modes | 10 min |
| Exploratory | Superpowers: TDD, Debugging, Code Review | Optional |
Key Concept: Context Engineering at Scale
Module 1 showed you what happens when you give an AI agent structured context for a CloudWatch alarm. This module scales that concept:
- CLAUDE.md encodes system context once and injects it on every interaction
- GSD workflow structures the context for entire projects — requirements, decisions, and plans
- Cross-session memory carries context forward across sessions — the AI doesn't forget what you built last week
- Plan modes let you choose how much structure the task needs
The pattern repeats throughout the course: structured context produces expert output. This module teaches the tools that make context engineering repeatable.