An open-source, self-paced course that teaches non-technical people to ship a real, deployed product using AI coding tools — and recover when the AI gets it wrong.
This repository is the canonical source of truth for the course: plain markdown files arranged into modules, each made of lessons that follow a single, locked anatomy. The same content is also wrapped by a Next.js platform (built, in site/) that adds accounts, per-lesson progress tracking, and cohort schedules; once it deploys to Railway it will serve at https://shipyourfirstthing.com. The lessons are identical on both surfaces, and this repository stays canonical — on day one you only need a browser to read it. The hands-on modules use a separate, tiny starter workspace you copy for yourself (Khrafts/syft-starter); SETUP.md walks you through it.
The course is designed to be picked up cold. A learner who has never written production code should be able to open this course in a browser, follow SETUP.md to set up their own workspace, and work through the modules in order without needing a workshop, a video, or a person to answer questions live. The "and recover when the AI gets it wrong" part is the differentiator: the modules teach the durable AI-coding loop alongside the toolchain, so by the end you can both ship a thing and get unstuck when the model produces something broken.
- People who are comfortable using a computer
- People who have never written production code
- People who have used GitHub at most to view a page
- People who are curious about building real software but feel intimidated by code
- People who want to work at their own pace, without a class or workshop
- People who may or may not have budget for paid AI tools (the course supports both)
- Read
SETUP.mdand choose your cost path fromBUDGET.md. - Make your own copy of the starter workspace (github.com/Khrafts/syft-starter) and open a Codespace on it — full instructions in
SETUP.md. You read the course here; you build in your own copy. - Work through Module 0, then Module 1, then Module 2 onward — in order.
- When something breaks, check
COMMON-ISSUES.md. When a page doesn't match what you see, checkWHAT-CHANGED.md.
Modules
- Module 0 — Welcome
- Module 1 — Mental models
- Module 2 — The developer toolchain
- Module 3 — Working with an AI coding agent
- Module 3.5 — Reading code, just enough
- Module 4 — Designing & building the thread project Coming in later phases
- Module 5 — Operating the build Coming in later phases
- Module 6 — After it's live Coming in later phases
- Module 7 — Where to go from here Coming in later phases
Cross-cutting docs
- SETUP.md
- GLOSSARY.md
- CHEATSHEET.md
- COMMON-ISSUES.md
- BUDGET.md
- CONTRIBUTING.md
- WHAT-CHANGED.md
- VERSIONS.md
Templates
Licensing
- LICENSE (MIT)
- LICENSE-content (CC BY 4.0)
- LICENSING.md
This repository ships under a dual-license model: code under MIT (LICENSE) and course prose, exercises, and diagrams under Creative Commons Attribution 4.0 International (LICENSE-content). See LICENSING.md for which files fall under which license.
See CONTRIBUTING.md.
AI tools change every few months — model versions, CLI flags, default behavior, even what the tools are called. The course's loop (intent → ask → evaluate → steer) is durable; the keystrokes are not. If reality drifts from a page, check WHAT-CHANGED.md before you assume the lesson is wrong.