2.0 AI-Driven Development Foundations
Overview
This module lays the groundwork for AI-native coding. You'll learn how to translate ideas into specs and use AI to generate quality code. The focus is on frictionless onboarding into AI-driven work — setting up tools, learning how to "think in prompts," and embracing new workflows.
Core Themes:
- Introduction to AI pair-programming and prompt engineering
- Liatrio's Spec-Driven Development process
- Getting reps with an AI-native codebase
- Building your first reusable AI skill — a pre-SDD refinement step (the
refining-ideasskill) you'll carry into later modules
What You'll Do
You'll work through all the open issues in the Emerald Grove Pet Clinic repository. These issues are your "reps" — structured exercises to build muscle memory with AI-assisted development. Alongside the issue-fixing work, you'll build a refining-ideas skill in 2.2 Build Your First Skill — a reusable pre-SDD step you'll rely on when Module 3 shifts from defined issues to open-ended feature design.
Use the fork of the Emerald Grove Pet Clinic repository you created in 1.5 Fork the Example Repository. That fork is your personal workspace for all the exercises in this module.
You should already have:
- Your own fork of the Pet Clinic repository
- The issue set created in your repository
- Pre-commit hooks installed and verified
- AI code review configured on your PRs
Primary Capabilities You'll Develop
Spec-Driven Development Proficiency
Learn to use the 4-step SDD prompt sequence:
- Specification — Transform a feature idea into a detailed spec
- Task Breakdown — Break the spec into actionable tasks with proof artifacts
- Implementation — Execute tasks with AI doing the coding
- Validation — Verify implementation matches the specification
You'll practice writing clear, machine-readable specifications and letting the AI handle implementation details.
Prompt Engineering & AI Collaboration
Develop skill in crafting effective prompts and instructions for coding agents. Learn strategies to:
- Iteratively refine AI outputs
- Fix errors and course-correct
- Verify the AI's work
- Manage context effectively
Essentially, you become an "AI project manager" — directing the model and keeping it on track.
Version Control & Quality Checks
Even though AI writes the code, you remain responsible for quality:
- Use Git effectively with AI-generated commit messages
- Review AI contributions critically
- Use automated testing and validation from the SDD workflow
- Respond to AI code review feedback on your PRs
Exercises
Fix All Open Issues
The Pet Clinic repository has open issues waiting to be fixed. Your job:
- Pick up an issue
- Write a spec for the fix (SDD Step 1)
- Generate a task breakdown (SDD Step 2)
- Have AI implement the fix (SDD Step 3)
- Validate and get AI code review (SDD Step 4)
- Merge the PR
Repeat for all issues. This is where you build your reps.
Complete at least 5 issues singularly — one at a time, driving each through all four SDD steps yourself — before attempting any autonomous or parallel workflows. Doing the reps by hand is how you learn the primitives and guardrails that keep an agent on track and build the trust to know when and how to hand off to an autonomous workflow.
Record Your Demo
As you work through the issues, record a demo (5-10 minutes) that shows:
- The issues you fixed
- How you used AI to solve them
- Your approach and any techniques that worked well
Post it to #liatrio-forge-demos as one of your twice-weekly demos.
Deliverables
By the end of the Foundations Module, you should have:
| Artifact | Description |
|---|---|
| Spec docs | Markdown specifications for each issue — kept in your repo, no separate submission |
| Task breakdowns | How you decomposed each fix — kept in your repo |
| PR links | All merged PRs with AI code review — live in GitHub |
| Test output | Proof that all acceptance criteria passed — kept in your repo |
| Demo recording | 5-10 minute walkthrough of your approach — posted to #liatrio-forge-demos |
Everything except the demo recording just needs to be up to date in your repo and available if a facilitator asks to see it — there's no formal drop box.
Exit Criteria
You can move to the Features & Systems Module when you have:
Fixed all the open issues in the Pet Clinic with 99% AI-written code.
Your PRs should show AI code review feedback addressed, tests passing, and a demo recording posted to #liatrio-forge-demos explaining your approach.
If this is taking longer than ~10 hours, reach out for help.
Reflect on Your Workflow
Before moving on, run the /insights command in Claude Code. After working through all the issues in this module, you have real conversation history for it to analyze.
/insights reviews your recent sessions and surfaces actionable suggestions — things like prompting habits you could improve, features you're underusing, or workflow patterns that could be more efficient.
Take a few minutes to review the suggestions. Pick one or two to intentionally practice in the next module. This is one of the fastest ways to level up your AI collaboration skills.
What You'll Learn
By completing this module, you'll have experienced:
- Delivering features in minutes or hours instead of days
- The discipline of spec-driven development
- How to direct AI effectively (and when to start fresh)
- The feedback loop of AI code review
This establishes your baseline for accelerated flow.