Public benchmarks can tell you a great deal about AI model capability. They still may not tell you which model will perform best on the work you actually need done.
This final post in a 3-part series shows how to investigate that gap with a small personal evaluation: a representative task, controlled user-visible conditions, and a rubric written before testing. It also examines what happened when I applied that method to three model systems.
I built HTML Deck Studio because I was frustrated with the limitations of existing presentation tools like PowerPoint, Keynote, and Google Slides. They lock your content inside their proprietary formats, are difficult to version control, hard to automate, and limited in what they can do. I wanted a system that was built on open standards, was highly-configurable, and supported the way I work with AI agents.
In this post, I will explain how HTML Deck Studio works, why I built it the way I did, and how it can help you create better presentations.
When several AI agents work on the same project, the code changes are not the only thing that can collide. One agent may need Python 3.10, another may need Python 3.12, and a third may be testing a dependency upgrade. If they all share the same environment, the work quickly becomes harder to reason about.
Git worktrees and Dev Containers solve different parts of that problem. Worktrees separate the code. Dev Containers separate the runtime. Used together, they make parallel work much easier to manage. There is one catch, though: a worktree can stop behaving like a normal Git repository once it is opened inside a container. I built a template repository to solve that problem and make it easy to set up the pattern in your own projects. Know more about in the next sections.
Knowing that benchmark scores depend on their test setup is only the first step. The next challenge is recognizing what different benchmark families actually measure—and where their evidence ends.
This field guide maps six useful categories that appear in model announcements and provides a repeatable way to read them together. The categories are practical rather than exhaustive, and some benchmarks span more than one.
This guide will walk you through the process of reverting to an older version of Visual Studio Code Insiders if you encounter issues with the latest build. Whether you're experiencing crashes, broken extensions, or performance problems, this step-by-step guide will help you get back to a stable version while waiting for a fix.
Every major AI model release arrives with a scorecard. One model scores 87% on one test, 63% on another, and claims to outperform last month's leader on a third. The numbers look precise, but they rarely tell regular AI users what to do next.
This post explains what those scores actually measure, why the same model can receive dramatically different results, and why even a well-run benchmark may not predict performance on your work.
As part of my ongoing learning journey, I was looking for a Flashcard App that was simple, local-first, and backed by a plain text format. I couldn't find anything that fit the bill, so I built my own.
With my Markdown Flashcards, you can write flashcards in plain Markdown, and the app will render them into an interactive study interface. The app also manages metadata like difficulty and review dates, which are saved back into the same Markdown file.
I have never been a fan of PowerPoint or Google Slides. I find them clunky, slow, and a pain to maintain. I much prefer writing in Markdown - it's simple, fast, and version-control friendly. But how do you present Markdown content without exporting to PDF or building a separate PPT file?
So I built the Markdown Presentation Tool - a lightweight VS Code extension that turns your Markdown files into beautiful, interactive slide decks with a single click.
It is designed to present documentation, meeting notes, or architecture diagrams without the friction of external software.
PDFs have a decades-long history as malware vectors, yet most VS Code extensions treat them like any harmless document. Because VS Code lacks a built-in PDF reader, developers usually rely on third-party extensions that embed full browser engines or silently pull rendering libraries from external CDNs.
This quietly introduces a massive attack surface and unwanted outbound dependencies.
So I developed a VSCode extension that does only one thing, in a secure way: Reading PDF Files. Install it today from the VSCode Marketplace: Safe PDF Viewer.
I built Mermaid Slideshow - a VS Code extension that turns Mermaid diagrams into a distraction-free slideshow.
It also works with Cursor, Windsurf, Zed, amongst others.
I built it because VS Code's built-in markdown preview renders diagrams inline alongside text - small, non-interactive, and buried in content. I needed a way to pull diagrams out and present each one full-screen, so you can actually read them during reviews and presentations.