TurboWarp is an open source software project maintained by a small group of volunteers.
The project does not accept donations and has turned down very large acquisition and advertisement offers (>= $100,000). We work on this software because it's fun and fulfilling. Please help us keep it that way.
Good bug reports are the single best way to help us improve TurboWarp.
We are not mind readers. If something "crashed", please be more specific: Did you get an error message? What were you doing before the crash? Screenshots are a lot more helpful than you would expect.
You can open issues in any GitHub repository and it'll be seen by someone. We can't always respond or fix everything, but they do all get read. If you're not sure which repository, just open it in the scratch-gui one and it'll get moved to the right spot.
AI has significantly changed the nature of code submissions and pull requests.
TurboWarp does not have a no-AI policy. We use AI ourselves, often in ways that are not externally-visible. We are aware of the many criticisms against AI, but have made the decision that their value outweighs the downsides. You are welcome to disagree. We do not force anyone to use AI.
The bar for external code submissions (that is to say, any PR you open) has substantially changed. It is now trivial for anyone to ask their AI to vibecode a 1000 line PR implementing some new feature. Please understand that a 1000 line PR is no longer a gift. It's a burden. If we merge your pull request, we are now indefinitely forced to maintain code that you may not even understand yourself. We believe that a human is ultimately responsible for any code they contribute, regardless of if the AI wrote it.
Additionally, we pay for access to the latest frontier models (again, we don't accept donations, so this is all out of pocket). We have the most experience with which agents understand our code best, write the most sustainable code, and have the most tooling and experience available to verify changes. Most contributions we see were made with low-cost models on free tiers. The gap between these models and the frontier one we use is vast.
The takeaway here is that if you have a new feature you want, please talk to us first. Convince us that the feature is good and interesting. You can - and probably should - go and make something with it so that we can agree it's a good idea. We'll work with you to figure out if we want it, whether we want you to code it, or if we'll figure it out ourselves in some amount. Most large PRs that get merged have historically received substantial modification by maintainers anyway.
Thanks for reading this. If you mention that you read and understand CONTRIBUTING.md in your pull requests, that might help us prioritize your change a tiny bit. Most people don't read this far...