About AI Kit Guides

Why I Write This Site

AI Kit Guides is an independent blog by Peng Zhou, a Java engineer. I write about using AI tools in software work: understanding code, reviewing changes, running tests, and investigating failures that a successful build can hide.

The aim is to give readers useful commands, examples, and the reasoning needed to check a result in their own environment.

What I Cover

I cover the AI tools people actually use — ChatGPT, Claude, Cursor, Codex, Copilot — and what goes wrong once you get past the demo:

• AI coding agents — Claude Code, OpenAI Codex, Cursor and GitHub Copilot, including setup, repository context and verification
• Spring Boot 4 migration — Jakarta EE 11, Security 7, Jackson 3, and the failures that compile clean
• Testing & CI — JUnit 6, Maven, Gradle, ArchUnit, Testcontainers and CI gates
• AI API limits — 429s, quota tiers and retry strategy for OpenAI-compatible endpoints
• AI fundamentals — model comparisons, prompt technique and hallucination control

Reading the Evidence

A documented product feature, an illustrative example, and an observation from a particular run provide different kinds of evidence. Check which one supports a claim before applying it to your own project. A single agent run does not establish a general success rate.

Earlier drafts included accounts of AI-created lab experiments without retained source projects or raw logs. Those accounts do not establish personal production experience or reproducible benchmarks. The October 2026 review removed unsupported measurements and recast those articles as technical explanations or explicitly illustrative procedures.

AI assistance is used in drafting and editing this site. It does not substitute for source checking or execution evidence. Synthetic fixtures and proposed evaluation procedures should be identified as examples, and must not be presented as measurements from a real run.

Where a new example has been executed, its evidence is limited to the stated environment and checks. The Spring Boot verification lab includes two version profiles, real HTTP and MockMvc checks, H2 migration exercises, source digests and an October 23, 2026 run record. The smaller Java verification source package demonstrates plain Java checks. Neither reproduces the site's historical agent experiments.

Corrections

Substantive revisions carry an updated date. Correction notes explain material changes to technical claims or the evidence behind them. To report an error, please send the article URL, the disputed statement, and a source or reproduction through the contact page.

About the Author

Hi, I'm Peng Zhou. I'm a senior Java engineer who's been building software for over a decade — and like a lot of developers, I got curious when ChatGPT launched in late 2022. By early 2023, I was using it every day: drafting emails, debugging tricky code, and experimenting with API integrations that actually made my work faster.

This blog focuses on practical questions about AI-assisted development: how to scope a change, review generated code, and establish what a successful check really proves.

I'm not affiliated with any AI company. Product descriptions, source-based comparisons, and executed examples are labeled according to their evidence. A comparison without recorded trials is not presented as a measured ranking.

When I'm not writing about AI tools, I build and ship my own side projects — and I document the unglamorous parts (deploying to Cloudflare Pages, setting up payments, going from prototype to production) over at ShipToCash.

You can find me on GitHub, where I share code from my experiments with AI APIs and automation workflows. If you have a question or a tool you think I should check out, reach out via the contact page — I read every message.

Get in Touch

Have a question, suggestion, or correction? I would like to hear from you. Visit our contact page to reach out.

Subscribe to our RSS feed for the latest updates.