Programming with AI
ENRU
A handbook by Lunario

Programming with AI

How to give an AI a task, read the code it wrote, and check what came out. For people who already write code together with an AI and don't want to take it on trust.

Start with 1.1Download PDF 4.2 MB

Russian edition, PDF

Ask
a request with context, a goal, constraints and a definition of done — so the AI builds what you meant, not what it guessed.
Read
files, functions, tracebacks and diffs well enough to tell what the code does and where it will break.
Check
run it, test it, catch invented methods and silent errors, and know when to roll back.

How it reads

113 short articles, each built the same way: what it is, what it looks like in code, what to tell the AI, what to look for, and a cheat sheet. Read in order, or jump to the one you need today.

Next to the text is a console: the JavaScript and Python examples run right in your browser and are compared with the output recorded in the book.

Contents

Foundations

1Mental model and working cycle
  1. 1.1Who is in charge here
  2. 1.2The working cycle
  3. 1.3What AI does well and what it does badly
  4. 1.4Three skills
  5. 1.5The cost of a mistake
  6. 1.6The illusion of finished code
2Reading code
  1. 2.1Why read code if the AI writes it
  2. 2.2Anatomy of a file
  3. 2.3Anatomy of a project
  4. 2.4Variables, types, data structures
  5. 2.5Functions
  6. 2.6Conditions and loops
  7. 2.7Classes and objects
  8. 2.8Exceptions and error handling
  9. 2.9Decorators and “magic”
  10. 2.10Asynchrony
  11. 2.11Dependencies
  12. 2.12Reading a traceback
  13. 2.13Reading a diff
  14. 2.14Red flags in generated code
3Stating the task
  1. 3.1Why the AI does the wrong thing
  2. 3.2Anatomy of a prompt
  3. 3.3Levels of detail
  4. 3.4Template — a new function
  5. 3.5Template — fix a bug
  6. 3.6Template — refactoring
  7. 3.7Template — explain this code
  8. 3.8Template — review this
  9. 3.9Template — write tests
  10. 3.10Technique — plan first, code second
  11. 3.11Technique — let the AI ask the questions
  12. 3.12Technique — input → output examples
  13. 3.13Technique — limit the blast radius
  14. 3.14The project context file
  15. 3.15How to write a mini-spec
  16. 3.16Prompt antipatterns
4Checking and debugging
  1. 4.1The rule of trust
  2. 4.2Running it and reading the logs
  3. 4.3Preparing an error to be fixed
  4. 4.4Narrowing the search
  5. 4.5Debugging with print
  6. 4.6Tests — read them and run them
  7. 4.7The acceptance checklist
  8. 4.8Hallucinations
  9. 4.9The "fixed one, broke another" loop
  10. 4.10When to stop and roll back
5Process, tools, safety
  1. 5.1Git as insurance
  2. 5.2Chat or terminal agent
  3. 5.3Session context
  4. 5.4When to start a new session
  5. 5.5Choosing a model for the task
  6. 5.6The environment
  7. 5.7Secrets
  8. 5.8Basic deployment

Stacks

6Python and application logic
  1. 6.1Anatomy of a Python project
  2. 6.2Idioms the AI writes constantly
  3. 6.3Typing, seriously
  4. 6.4Asynchrony in practice
  5. 6.5Stating tasks in Python
  6. 6.6AI mistakes in Python, and a checklist
7Web backend
  1. 7.1How a backend is built
  2. 7.2HTTP in plain terms
  3. 7.3APIs, schemas and validation
  4. 7.4Reading backend code
  5. 7.5The database
  6. 7.6Queries that drag
  7. 7.7Migrations
  8. 7.8Access, AI mistakes, and a checklist
8Frontend
  1. 8.1How a browser assembles a page
  2. 8.2Reading HTML
  3. 8.3Reading CSS
  4. 8.4Developer tools
  5. 8.5AI mistakes in the frontend, and a checklist
9JavaScript and TypeScript
  1. 9.1Reading JavaScript
  2. 9.2Asynchrony in JavaScript
  3. 9.3What TypeScript adds
  4. 9.4How to read other people's types
  5. 9.5Types as a check on the AI's work
  6. 9.6AI mistakes in JS and TS, and a checklist
10React in TypeScript
  1. 10.1The anatomy of a React project
  2. 10.2Components and JSX
  3. 10.3Props and state
  4. 10.4Effects
  5. 10.5Stating interface tasks
  6. 10.6AI mistakes in React, and a checklist
11Flutter and Dart
  1. 11.1The anatomy of a Flutter project
  2. 11.2Dart in half an hour
  3. 11.3Widgets
  4. 11.4State
  5. 11.5Stating mobile interface tasks
  6. 11.6AI mistakes in Flutter, and a checklist
12Building and shipping
  1. 12.1What a build is and why it breaks
  2. 12.2Android and iOS
  3. 12.3Shipping and updates
  4. 12.4A checklist before shipping

Cross-cutting topics

13Architecture in plain terms
  1. 13.1How to break an idea into parts
  2. 13.2Folder structure
  3. 13.3Where to keep the data
  4. 13.4Configs and environments
  5. 13.5How to describe the architecture to an AI
  6. 13.6Questions before starting a project
14Workshop
  1. 14.1Case — a Telegram bot with an external API
  2. 14.2Case — a web API with a database
  3. 14.3Case — a React site for that API
  4. 14.4Case — a mobile application
  5. 14.5Case — an evening's script
—Appendices
  1. ARequest templates
  2. BA code review checklist
  3. CCommon Python errors
  4. DCommon JS, TypeScript and Dart errors
  5. EGit — rescue scenarios
  6. FThe terminal — the necessary minimum
  7. GGlossary

About this book