Write code with an AI partner.

Git for beginners: 8 commands that save your project from AI mistakes

Git is the undo button for AI coding. Set it up once, then learn the eight commands that let you check, undo and recover anything a coding agent changes.

Sep 23, 20269 min read

Run a coding agent in GitHub Actions: reviews, fixes and triage

How to run a coding agent in GitHub Actions in 2026: the official options, a workflow file from the docs, safe permissions and secrets, and cost limits.

Spec-driven development: write the spec, let the agent build

Spec-driven development explained: what a good spec contains, a copyable template, the tools behind the trend, and how to keep it from becoming waterfall.

Sep 23, 20268 min read

How to refactor legacy code with AI without breaking it

Refactor legacy code with AI safely: map it first, pin its behavior with characterization tests, change it in small commits and prove nothing changed.

How to migrate a codebase with AI: upgrades, ports and rewrites

How to migrate code with AI: inventory the work, let codemods do the mechanical part, test every batch, and learn from Airbnb, Slack, Google and Stripe.

AI code review tools compared (2026)

The AI code review tools that matter in September 2026, checked on their own pages: what each reviews, what it costs, how to set it up and how to trial it.

AGENTS.md explained: the file your coding agents read first

AGENTS.md is a plain Markdown file that tells coding agents how your project works. Which tools read it, what to put in it, and a complete example.

What is vibe coding? How it works and where it breaks

Vibe coding is building software from AI prompts without reading the code. Where the term came from, what it suits, where it breaks and how to do it safely.

The best AI coding agents in 2026, compared

Claude Code, Codex, Cursor, Copilot, Antigravity, Devin, Kiro and more: what each AI coding agent does best, what it costs and what changed in 2026.

Claude Code vs Codex vs Cursor: which one fits how you work

A fair 2026 comparison of Claude Code, OpenAI Codex and Cursor: interfaces, models, prices, strengths and weak spots, with a decision table and three scenarios.

Should you still learn to code in 2026? A straight answer

Yes, but not the way people learned five years ago. What the 2026 data says about AI and developers, the skills that matter now, and a 90-day plan.

12 prompts for coding agents that get working code

Twelve copyable prompts for coding agents, from planning and tests first to security reviews and commit messages. Each one says when to use it and why it works.

The AI pair programming workflow that actually ships

A practical AI pair programming loop: plan, one small step, run, review, commit. Who does what, how to give context, and when to take the keyboard back.

TDD with AI: let the tests keep your coding agent honest

TDD with AI, step by step: write the tests first, lock them, and let the agent code until they pass. A worked Vitest example and how to stop test tampering.

How to debug code with AI: a step-by-step method

A step-by-step method to debug code with AI: reproduce the bug, gather evidence, test one hypothesis at a time, bisect, fix the cause and add a regression test.

How to review AI-written code without reading every line

A practical method for reviewing code an AI agent wrote: check intent first, run it, triage by risk, and let tools read the boring parts. Checklist included.