Aikarai Learn · start small
From zero to “AI, karai” in five steps.
1. Ask. 2. Test. 3. Publish. A practical path from a two-hour idea to a tested local project, a clean source package, and a truthful Aikarai publication.
Learn pages are public and anonymous. They do not create an account, a session, analytics, or marketplace activity.
Lesson 01Turn one small annoyance into a local tool with a visible result and a short, testable definition of done.
- Suggested time
- 20 minutes to scope · 1–2 hours to build
- You will finish with
- A narrow plan, a runnable check, and no mystery features.
Start this lesson →Lesson 02Give Codex a crisp goal, enough repository context, hard constraints, and an observable definition of done.
- Suggested time
- 15 minutes to set up · 60–120 minutes to build
- You will finish with
- A reviewed implementation with commands and browser evidence, not a vague completion claim.
Start this lesson →Lesson 03Use the gather-context, act, and verify loop, with a runnable check that lets the agent correct its own work.
- Suggested time
- 15 minutes to set up · 60–120 minutes to build
- You will finish with
- A small implementation verified against an explicit user journey.
Start this lesson →Lesson 04Create a minimal, clean archive with setup instructions, disclosure, limitations, and reproducible verification.
- Suggested time
- 30–45 minutes
- You will finish with
- A source ZIP that is useful after download and safe to inspect.
Start this lesson →Lesson 05Use Aikarai Studio to freeze a truthful snapshot, process the package, publish a lane, and verify the public and claimant journeys.
- Suggested time
- 45–60 minutes
- You will finish with
- A public Build whose claims, package, lane, and Launch match what you actually tested.
Start this lesson →