Frequently asked questions

I've already signed in with GitHub - why do I need to connect with GitHub again?

Signing in provides basic GitHub profile details like your profile picture and email address which we use for authentication. Connecting a second time will request read-only private repository access - but please note this is totally optional and you can still use the site without it.

The GitHub integration wants read-only repository access. Will you access my code?

Absolutely not. We only use the GitHub integration to read commit history - it would actually be far better for us if GitHub offered more fine-grained access to request only that. Nevertheless, your code will never be accessed.

I don't want to grant private repository access. Can I still use the site?

Yes! You can still use all the tools with public GitHub repositories or by manually entering commit history.

Can I try any of the tools for free?

Absolutely! You need to sign in with GitHub first so we can track credit usage, but when you sign up you'll get 5 free credits to play around with each tool.

How does pricing work?

Digest Diff is pay-as-you-go. You buy credits, and those credits can be used on any of the tools. The more credits you buy upfront, the cheaper they are.

What does a credit get me?

Release notes and recaps are fixed at one credit per use. Codebase overviews let you choose between generating a basic summary (one credit) or detailed history (two credits).

Credits seem to cost more than other AI tools. Why is that?

Digest Diff makes exclusive and very extensive use of commit history. This typically involves sending a lot of 'context' to ChatGPT, and more context unfortunately costs more money. We're looking at ways we can make credit usage dynamic based on commit history to mitigate this.

Is this just ChatGPT under the hood?

Yes and no. The model at the heart of the system is indeed OpenAI's GPT-4, but a vast amount of effort has gone into choosing the right prompts for each tool, making those prompts adapt to the numerous choices you can make, not to mention fetching and squeezing as much commit history into the allowed context window for GPT-4 as possible.

Do you store or inspect any generated recaps, overviews or release notes?

Absolutely not. Responses are streamed directly to your browser: we do not log them nor store them. You may choose to export them once generated, but that's entirely up to you.

My question wasn't answered here - how can I get in touch?

Please use the Feedback navigation link and be sure to leave a return email address if you'd like a response. Alternatively follow the footer link to find various ways of contacting me.