Theme: Engineering with AI, Not Around AI — the program capstone. The draft → verify → log loop you've run through every module becomes the explicit subject, made rigorous: how a team engineers Project Forge with AI, safely, at scale, with the engineer accountable for every line.
You are an AI Software Engineer. AI can now draft large amounts of Forge's code in seconds — an opportunity and a liability. Across 9 ticket-driven lessons you build the Forge AI engineering platform: context engineering so AI works from how Forge really is, architecture-aware prompting enforced by a fitness function, AI-assisted TDD (test first, AI second), rigorous review of every line, fearless refactoring under characterization tests, gated and scoped autonomous workflows with a human checkpoint, and governance for secrets, prompt injection, and dependencies. The bottleneck shifts from writing code to verifying it — and verification is the skill this module builds.
The ethos, in every lesson: engineering with AI, not around it; you are accountable for every line ("the AI wrote it" is no defense); context is the program; test first, AI second; verify, don't trust; AI changes the cost of code, not the standards for it; govern the boundary; and automate with guardrails. The whole module runs on draft → verify → log.
| Step | What it means |
|---|---|
| Pick up a ticket | Each lesson is an engineering ticket (CTX-1010, GOV-4010, …) with acceptance criteria |
| Engineer the context | Give AI the types, signatures, constraints, and examples it needs |
| Draft with AI | AI produces a focused artifact (code, test, refactor, plan) |
| Verify against something you control | a failing test, a fitness function, a review, a gate |
| Log the exchange | what you asked, how you verified, what you kept or changed |
| Vouch for every line | approval means you understand it and own it |
By the end you will be able to:
- Run the draft → verify → log loop and keep an AI-usage log, accountable for every line.
- Engineer context (types, signatures, constraints, examples) so AI output fits the system.
- Enforce architecture with a fitness function that fails AI code crossing a boundary.
- Use AI-assisted TDD: write the failing test first, have AI implement to green.
- Review AI code rigorously and turn every finding into a regression test.
- Refactor under characterization tests so behavior changes are caught.
- Orchestrate gated, scoped autonomous workflows with a human checkpoint.
- Govern AI usage: block secrets, treat untrusted content as data, gate dependencies.
| # | Lesson | Competency | Ticket |
|---|---|---|---|
| 0 | Welcome to the AI Engineering Team | AI Engineering Orientation | AIE-1000 |
| 1 | AI Is Your New Pair Programmer | Context Engineering | CTX-1010 |
| 2 | Make AI Follow Your Architecture | Architecture-Aware Prompting | ARCH-2001 |
| 3 | Test First. AI Second. | AI-Assisted TDD | TDD-2010 |
| 4 | Review Every Line | AI Code Review | REV-3001 |
| 5 | Refactor Without Fear | AI Refactoring | RFCT-3010 |
| 6 | Orchestrate Autonomous Engineering Workflows | AI Workflow Orchestration | ORCH-4001 |
| 7 | Protect the Engineering Organization | AI Security & Governance | GOV-4010 |
| 8 | AI Engineering Platform | AI Engineering Platform | FORGE-9800 |
Phases: Foundations (0) → Working With AI (1–2) → Verification-Driven AI (3–5) → Scaling & Governance (6–7) → Capstone (8).
.
├── README.md # this file
├── MODULE_SYLLABUS.md # pacing, structure, deliverables
├── LEARNER_GUIDE.md # how to operate as an AI software engineer here
├── INSTRUCTOR_GUIDE.md # facilitation and assessment
├── COMPETENCY_MATRIX.md # lesson → competency → skills
├── ASSESSMENT_RUBRIC.md # grading weights and performance levels
├── dashboard.html # interactive progress dashboard (open in a browser)
├── Lesson_00.md … Lesson_08.md # the 9 lessons
├── labs/ # hands-on labs (+ aitools.mjs helper; run real node:test checks)
├── solutions/ # worked solutions / answer keys
├── resources/ # context, architecture, TDD, review, refactoring, orchestration, governance + more
├── assignments/ # submission templates + capstone brief
└── instructor-notes/ # per-lesson facilitation notes
- Read
resources/ai-engineering-setup-guide.md; the runner is built in andaitools.mjsships inlabs/(Lesson 0 /labs/lab-00-setup.md). - Start your engineering notebook from
resources/engineering-notebook-template.md— including the AI-usage log. - Open
dashboard.htmlin your browser to track progress through the lessons and phases. - Open
Lesson_00.mdand pick up your first ticket. Keep the relevantresources/references open as you build.
Verification. This module is about turning AI drafts into trustworthy software, so verification means running real checks: the built-in node:test runner executes actual suites with real # pass/# fail counts and exit codes — AI-assisted TDD shows red before green, review shows a planted bug caught, refactoring shows characterization tests staying green; architecture conformance is an executable fitness function, orchestration a gated workflow, and governance a set of policy functions (secret scan, injection detection, dependency gate), all asserted with node:test against the shipped aitools.mjs helper. The point of every check is that you can vouch for the AI-generated code as if you wrote it.