Jazzberry is an AI agent for finding bugs that is driven by real code execution. You can add Jazzberry on GitHub to start automatically getting bug reports on your pull requests.
We're Mateo and Marco and we've built Jazzberry, an AI agent for finding bugs. Jazzberry is powered by running your code in a secure sandbox and integrates directly into your pull request workflow on GitHub.
When a PR is made, Jazzberry starts by cloning your repository into a secure sandbox. Our agent then examines the contents of the pull request to guide its search for bugs. In the sandbox it can write and execute your code much like a person would. When Jazzberry is done, it returns a markdown table as a comment on your PR.
You can check it out at jazzberry.ai. Looking forward to the PH community giving it a try!
This looks interesting, we are currently using CodeRabbit and we did notice a few things missing, like custom rules. Will Jazzberry consider this as part of product milestone? @garrytan
How often does it misidentify a bug? Have you had cases where it is wrong?
Jazzberry found bugs in the airweave codebase and made us feel insecure, but Marco and Mateo were nice about it 👍
Congrats! Would love to see if this could work with an automatic Claude Code workflow where it automatically "pushes back" against some non-PR changes, so Claude can fix issues before "finishing".
congrats on the launch guys super cool! If code is poetry, Jazzberry is a literary critic.
There’s something really charming about Jazzberry—it doesn’t just feel like another AI tool, it feels like a creative partner with a bit of soul. Love the playful vibe and the care that clearly went into it. Congrats to the team, you’ve made something special here
So cool! How do you plan to handle potential false positives as the system evolves?
Congratulations on launching! Very talented team.
Question - how does it handle git submodules? I might be overthinking it, but will it also install Jazzberry on those repos, or will the agent know to git install with submodules and then do its thing?
Jazzberry sounds like a fantastic tool for streamlining the bug-finding process, Mateo and Marco! Integrating directly into GitHub PR workflows is a great touch, making things so much easier for developers. Excited to see how this enhances code quality. Congrats on the launch! 🚀
Absolutely love this idea! This is so much more important now that AI is writing all the code!
I really like the pragmatic angle: instead of “AI pair-programmer” hype, you’re focusing on the unglamorous (but critical) job of catching bugs before they land in main. An this is amazing.
Two ideas and a question as I read the docs:
Signal-to-noise tuning: will teams be able to set severity thresholds? (e.g., fail the check on crashes or data loss, but only comment on style nits). Nothing seems to kill adoption faster than alert fatigue.
Custom test seeds: exposing a YAML file where devs can define typical env vars or sample payloads would let Jazzberry hit the weird edge cases we see in prod.
Language roadmap: today it looks Python & JS-heavy; any ETA on JVM stacks? A lot of fintechs seems to be still deep in Kotlin/Java land.
Anyway, great launch: automated bug hunts have been overdue for an upgrade! 🚀
Love the focus on security issues like SQL injection and authentication bypass. Finding those early can save so much headache and risk.
Jazzberry is such a smart and helpful tool—love how it runs real code to find bugs automatically. You should totally list it on Aixyz .Co ! It’s free, and a great way for more devs to discover tools like this. Keep it going!
Big shoutout to Jazzberry — such a clever little bug detective! It’s adorable how it runs real code to catch issues and delivers automatic bug reports right on your GitHub pull requests. Debugging just got a lot more fun and stress-free!
