Over the past year, I’ve shipped features and entire applications using Cursor, Claude Code, and Codex. Like everyone else who’s made the switch, the speed of shipping blew me away initially (and it still impresses me to this day) but as much as I enjoy the process and the boost in productity, there was this voice in the back of my head that wondered if the speed was something that I always optimized for, back when manually coding, or something that effectively agents allowed me to do now and that I leaned into heavily? After spending years in management before returning to hands-on building, I kept wondering: has agentic coding actually changed how I approach building software?
Not the tools. Not the speed. Me. I know my type. I was never the TDD-first, spec-everything-out developer. I was the “get to a prototype quick and learn from it” type. And I think a lot of programmers share a similar honesty, we knew about best practices, better architectural patterns, lean code, and all the methodologies that would help us ship better code. But company culture, time pressure, and tech debt played heavily against our best instincts. Delivery is what mattered.
So I wondered: now that AI agents can bang out complex, optimized code in seconds, now that you can ship a production-ready app in a weekend if you know what you’re doing - does that change coding personalities?
One weekend I started building to find out and today I am excited to share the Vibe Coding Profiler:
https://www.vibe-coding-profiler.com
It started as Project BoloKonon (from the Bambara word meaning “inside the hand” — as in craft). I wanted to create a personality test for my (and your) vibe coding style. What I ended up with is a Vibe Coding Profile (VCP) built on 6 axes that came out of my research. The underlying design principle that drove it was:
What dimensions of “how you build” with coding agents are actually observable from git metadata alone?
This is because a thorough analysis would involved sharing part of all of the prompts used as part of the building process, which is both impractical and unsafe so I wondered if git history could (pun intended) tell a story about my vibe coding process. The 6 axes (A through F) used in generating the profile are a synthesis of various existing research or writings:
TDD and Agile academic research: test-first patterns and planning signals, informing Guardrail Strength (B) and Planning Signal (D)
GitClear 's developer productivity studies informed code churn and commit pattern analysis, informing Iteration Loop Intensity © and Shipping Rhythm (F)
Addy Osmani’s excellent writings on agentic coding, especially the conductor vs. orchestrator distinction, which directly informed axes like Automation Heaviness (A) and Surface Area ()E
A persona taxonomy synthesis from multiple industry frameworks
How it works:
Connect with GitHub (GitLab and Bitbucket also available)
Connect a repository
VCP runs an analysis on the metadata from a sampling of commits in your git history
You get assigned a persona based on your patterns
You can optionally enable an LLM-powered analysis for a more comprehensive review of your methodology. Each repo you analyze feeds into a Unified VCP: your baseline coding profile across projects. Since your style might shift depending on the project, you get both individual insights and the aggregate view.
You can also choose to share your VCP publicly, mine is here:
https://www.vibe-coding-profiler.com/u/devakone
Your profile is private by default, and you choose how much to share. I took security and privacy seriously but it goes without saying, use this for side projects, not work repos. The project is open source and you can contribute at
https://github.com/devakone/vibe-coding-profiler
I’ll keep refining the signals as the industry evolves and hopefully others have better idea and as my brother Yannick suggested, maybe this could evolve in a diagnostic tool to help early career coders improve their approach to building effectively. There was quite a bit of iteration, but I’m glad to release V0 today and let the power of community and open source help evolve it
Give it a vibe and your feedback is always welcome!
Top comments (6)
I have been seeing people talk about vibe coding all week and it is such a cool concept to explore. I am really curious about how you came up with the different personality types for the test. As a student still figuring out my own coding style it is fun to see where I might fit in. Did you build this using a specific framework or was it just a creative side project? I would love to hear more about how you mapped the traits to the results.
This was a creative side project because I wanted to answer that question for myself as well! I did some (of course AI assisted) research on the topic, it is still new and there aren't quite any industry standards yet. The metrics I came up with are an aggregate of various research and opinions. Most of all I wanted to build something that would resonate with me and help me make sense to how I approach vibe coding.
okay thats fair enough
ha this is great. took the quiz and got "The Pragmatic Prompter" which honestly tracks - I do spend way too much time tweaking prompts before letting the AI run. been doing a lot of vibe coding lately with a security scanner I built and the biggest lesson is that verification > generation every time. cool project!
Thanks Mykola! It's gratifying to see that it resonated with you and was accurate enough to capture your actual flow!
honestly the categories are pretty spot on. I've noticed I switch between modes depending on what I'm building - quick prototype? full prompt mode. security-sensitive stuff? way more hands-on. clever framing tbh