What does Rapid Review send to AI model providers?
To review a pull request, Rapid Review sends AI model providers:- the changed code of the files it reviews, with the surrounding and related code the review’s mode includes, which can come from elsewhere in your repository and from packages your code depends on;
- the pull request’s title, description and linked issues, to work out what the pull request says it does;
- in repositories that record learnings, the replies in Rapid Review’s own review threads.
.rapid-review.yml excludes, aren’t sent.
Is my code used to train AI models?
No. We send review requests only to model providers that do not train on the data.Do model providers keep my code?
We don’t require zero data retention. So a model provider may keep request data, including your code, under its own policy for a time. We send review requests only to model providers that do not train on the data.Which files are never sent to a model?
Changes to files that look like secrets, such as.env files, private keys and credentials, are never sent to a model, whatever the repository’s include rules say. The check lists them under Warnings.
Files your .rapid-review.yml leaves out aren’t read for the review either, also when the reviewer looks at other code in the repository.
What does Rapid Review store?
Rapid Review keeps:- a record of each review and its findings;
- the summary of each reviewed pull request;
- the learnings, and a history of changes to learnings and review settings;
- each reviewed pull request’s title, author and state.