# Explaining Unfamiliar Code — AI Coding Assistants: Use Them Well and Review Them Hard

Source: https://www.geekswithgeeks.com/en/ai-coding-assistants/flow-explain

> Use an assistant as a reading companion for code you did not write.

## Ask for layers of explanation

When you join a new codebase, ask the assistant for a **map first** ("what are the main modules and how do they connect?"), then drill into one file, then one function, then one tricky line. Ask it to explain **why**, not only what, and to give an example input and output. Then verify the claims against the code, because a confident explanation can still be wrong.

## Understand, build, fix, improve

Four tasks cover most daily use: explaining code, writing it, debugging it and refactoring it.

![Four tasks: explain, write, debug, refactor.](assets/figures/ai-coding-assistants/section-3-map.svg) — Figure 3.1 — Explain, write, debug and refactor.

## A prompt for reading code

Asking for a worked example makes the explanation concrete and easy to check by running it.

```text
Explain this function for a new team member.
1. One sentence: what it does.
2. Walk through it line by line, noting any non-obvious parts.
3. Show a sample input and the exact output.
4. List edge cases where it could fail.

<paste the function>
```

## Run it to confirm

If the explanation says "returns the last three items", run the function on a sample and see. Turning claims into small experiments is the fastest way to learn and to catch wrong explanations.

**Quiz:** How should you treat an assistant's explanation of code?

- [ ] As useless
- [ ] As always correct
- [x] As a helpful hypothesis to verify against the code
- [ ] As a legal document

*Answer:* As a helpful hypothesis to verify against the code. Explanations speed up understanding, but you confirm them by reading or running the code.