Blogmark

From Justin Searls:

It's been a weird experience enabling turbocommit on my repos and watching it do a better job titling commits than I ever do. That it preserves my agent transcripts in the commit message alongside code changes is really nice!

In particular a tool like this feels like a win for engineering organizations that cannot convince all their developers to write meaningful and useful commit messages. Instead of, "fixed bug", you get the context that was embedded in the prompt. Whatever prompt and subsequent convo was needed to give sufficient context to the LLM will be captured as a version control artifact.

Even as I tend to put effort into my commit messages, the process of summarizing what it took to get to the fix is fairly lossy. Automatically capturing all the details that I think will be useful to the LLM in commit messages helps preserve useful detail.

This also plays into this idea of context synergy that I've been thinking a lot about.