Baby Dev's First Commit

Mason James

Apr 6, 2025 — 2 min read

After a year's break, I'm writing again. The reason?

I successfully made my first commit to fix a bug in a cool community repo I've used for years. This isn't so exceptional at first glance, but I'm not a coder. I don't write javascript or python or php.

Last week, while testing the incredible Claude Code (thanks again for the swag btw) I had the idea of creating a highly-specific commit to a project I'm unfamiliar with. Could I direct Claude Code to a bug, have Claude fix it and then submit a pull request (code patch) that would be accepted by the project maintainers?

Well, as I alluded to above, yes it can. I opened up one of my favorite open-source projects, TubeArchivist, literally clicked on the most recently reported bugs and passed that to Claude Code. It took Claude only seconds to scrape the entire codebase, diagnose the issue, and supply a small code fix. The total cost was $0.56 in API tokens. (Most of this time was creating a thorough CLAUDE.md file to provide the codebase context. This is mostly a one-time cost.)

I wanted this to be a real test of Claude, and also, not have someone put off by AI-generated code, so I manually created a branch and PR request with Claude Code's advice. I used the Github desktop app for this as I also do not really get git. Anyway, I submitted it to the wrong branch, but the maintainers kindly moved it, realized which bug it was fixing, and scheduled the change to be included in core.

All good to this point. In fact, when I went back to the repository to check how things were going, the code fix has been referenced on other issues as a resolution. One thread has garnered more than 100 comments and the Claude Code commit gets a reference. So should we all install Claude Code and begin solving the bugs in our lives? Well, maybe, but there's a real case for caution. We've seen this playout a few times already with AI. First with art shows, then novel-writing competitions get overwhelmed. Wikipedia can't keep up with the AI-assisted article updates, and there's reason to believe open source project maintainers will face the same thing.

We'd do well to think about the potential flood of AI-generated slop disguised as code contributions. Though, I imagine we'll shortly have AI's reviewing the code commits.

REPO PROMPT REVIEW: Granular Control and Workflow Excellence

This is the fourth post in my ongoing series of exploring AI-assisted coding ahead of o3 Pro and o4. Check out the initial overview, best practices for AI coding, and my first review on Cursor and Windsurf. Repo Prompt - might seem like the dark horse candidate here, but I&

CURSOR REVIEW: Debugging with AI Precision

Cursor rightfully leads the AI-assisted coding space. I use it daily, particularly when debugging active web apps built with frameworks like Next.js. Cursor's strength lies in terminal interactions, drastically reducing debugging time. A notable project was a Next.js + TailwindCSS + Shadcn UI admin dashboard for an AI

BEST PRACTICES: Key Understandings for AI-Assisted Coding

This article is part two in my AI coding series ahead of this week's anticipated pre-o4 model announcement. In the first article I looked at the overall AI coding landscape and why it's exciting. I also have specific app reviews for Cursor and Repo Prompt with

The Apps for Coding before o3 (April 2025)

In anticipation of OpenAI's release of the full o3 model later this week. I thought I'd take a few minutes to reflect on the current state of coding with AI and mention a few of the practical solutions I've been testing (and had success

Read more

REPO PROMPT REVIEW: Granular Control and Workflow Excellence

CURSOR REVIEW: Debugging with AI Precision

BEST PRACTICES: Key Understandings for AI-Assisted Coding

The Apps for Coding before o3 (April 2025)