Learn to work with AI coding agents through short, interactive lessons
You ask a coding agent to fix a feature. It edits the files and says it's done. What do you check before trusting the result?
I'm building Kodivo to help people learn how to work with AI coding agents. It's free, runs in your browser, and uses short, interactive lessons with examples and practice questions.
The AI Coding path covers giving agents clear tasks, choosing the context they need, and checking their work. You can try a lesson without creating an account.
Here's an example of what you'll practice.
Turn a vague request into something you can verify
Consider this task:
Fix the notes feature. The agent has to work out what needs fixing. Is the problem with saving a note, displaying it, or editing an existing one?
A more specific request describes the behavior you want:
When I save a note on a task, the text should still be there after I leave the task and reopen it. Investigate why the note disappears and fix that behavior. You now have a concrete sequence to try yourself. Enter a note, save it, leave the task, and reopen it. Compare what you see with what you expected.
That also gives you a better way to assess the agent's response. You can look for evidence that it checked the behavior you described.
A working-looking button can still lose your data
Kodivo's public AI lesson follows an agent that changes the code behind a Save button and reports that it works.
You save a note and reopen the task. The note is empty.
A screenshot could show that the button looks correct. A list of edited files could show where the agent worked. To verify the requirement, you need to follow the note through saving and reopening.
A useful follow-up request would be:
Check that a saved note survives reopening. Report the expected note and the observed note. If you cannot run that check, say so. The agent might not have access to a running application. It might be able to inspect the code but unable to test the full interaction. Its response should make that limitation clear.
In the lesson, you work through this scenario and choose which evidence would actually check the requirement.
Practice the decisions you make while coding with AI
Working with a coding agent involves decisions throughout the task. You decide what outcome to ask for, which information to provide, and how to evaluate the proposed change.
Kodivo's AI Coding lessons let you practice those decisions using concrete scenarios.
Each lesson breaks the explanation into short cards. You read an example, move through the reasoning, then answer a question to check what you understood.
On your phone, swipe up to continue through the cards. Navigation controls are available too.
The format is meant to make it easier to work through a small part of a topic at your own pace. You can try an example before committing to an account.
Learn in your browser or connect your own agent
You can take the lessons entirely in your browser. You don't need an AI subscription or a connected agent to read the lessons and answer the questions.
If you prefer learning with your own tools, Kodivo also supports connecting Codex or Claude Code through MCP. A connected agent can access your Kodivo lessons and saved progress.
That optional connection uses your own provider account. Kodivo's free lessons don't include an external AI subscription or provider usage.
The lessons and interface are currently in English.
Try a free AI Coding lesson
Start with the Save-button example. Read the cards, follow the scenario, and try the question at the end.
Try the AI Coding lesson without signing up
Create a free account when you want to save your progress and continue learning.
Kodivo also has a separate PHP learning path if you want to study the language alongside your work with coding agents.
I'm the maker. If you try the AI lesson, tell me which explanation was useful or which card left you with an unanswered question.