Ever been confused by a block of AI-generated code? Gait is an open-source Cursor and VS Code extension that lets you view the prompt that created it. Continue off of your coworker's conversations, view analytics on your AI codegen use, and more!
Hi Product Hunt makers!
I’m Kyle, co-founder of Gait. Alex and I have spent the last couple weeks building gait. As heavy users of tools like Github Copilot (back in the day) and Cursor, we’ve been really excited about the power of these tools to build quick prototypes, demos, and learn new technologies. However, we’ve been frustrated by how hard it is to build larger applications with AI, especially with multiple people on the project.
How we help
AI code-gen tools are built for individuals, not teams, and code is read much more often than it is written. Increasingly, understanding and editing becomes the limiting factor in a team’s velocity, not generation.
🧐 AI-generated code is hard to understand
We use Cursor to ship quickly, but AI diffs are often destructive, large, and opaque. Prompts serve as automatic documentation.
📈 It’s hard to understand AI’s impact on your codebase
gait is the first layer of observability, giving oversight into you and your team’s use of AI and its impact on productivity. Take a look at the stats for our repo. See your own stats here.
How it works
gait will automatically read through your chat history and save the chats and edits you make to your codebase. This state is written into a .gait folder, which you can check in to share with your coworkers.
Features We Want to Build
Gait, as it is, helps humans understand code. We want to build out additional features that give LLMs additional context and understanding that make editing with AI more accurate and effective. We think the repo should contain all the context an AI agent or assistant needs to change the codebase. In addition to prompts, this includes things such as Google Docs, Slack, Jira, etc. but also other forms of context that are crucial. One thing we’re currently adding support for is “comments for your AI” - additional information that gets automatically added to context without having to clutter your file.
Our Ask
Try us out! We’re early and iterating on the product. We’d love to have your feedback and we welcome any contributions to our repo (we are open source).
Our repo: https://github.com/gait-ai/gait
Our discord: https://discord.com/invite/yKvt3...
Download and get started with gait: https://getgait.com/getting-started
If you're building any AI codegen tool, we’d love to integrate! We currently support Cursor and Github Copilot, but are keen to support more tools.
The integration of prompt tracking with generated code is a clever approach. It seems like a great way to enhance collaboration within teams.
Git blame for LLMs is a cool idea! I don't think I've seen anything like this before. This feels like a super natural evolution of version control.
This is insanely cool! Cannot tell you the number of times I’ve wished that this existed. Documentation with prompts is brilliant. Excited to try it out!
Amazing! Sometimes I think that it takes the same amount of time to code with an LLM as without, as I have to spend so much time debugging. Really glad to see a tool that helps explain code outputs, will try this in my workflow!
This tool is so cool. When working in a team, it is so useful to know what kind of code PR author has added to the codebase which was generated by an LLM. This can reduce errors and bring a fruitful discussion. Upvoted!
I can finally decipher what code my CTO has been writing. In all seriousness, really neat tool - excited to keep playing around with it!
Absolute game changer ... can immediately see this aligning my team of devs. Self documenting codegen is also a huge plus!! Super excited to use this
Hey Product Hunt!
Alex, co-founder of Gait here. We're really excited for you to try gait out! We believe that storing prompts and the corresponding code that's generated from them allows for all sorts of cool applications, which is why we chose to open-source gait!
Really excited to see what people are able to build on top of this!
If you have any ideas/feedback, feel free to email us at [email protected]
This is so cool! I’ve experienced this issue too. Version control needs improvements like this to keep codebases clean and maintainable.
I’m really excited to use this. Strongly support!
