It gets better at your repository
When a pull request closes, inline goes back and looks at what happened to each finding: the reaction it got, whether its thread was resolved, and whether the flagged code actually changed. That is a real outcome, not a thumbs-up nobody clicked.
Dismissals become
rules you approve
A pattern dismissed enough times becomes a proposed learning with the evidence attached. You accept it, edit it, or reject it. Accepted rules go into the next review; nothing changes behind your back.
Every charge goes through `src/charges.ts`; nothing calls the Stripe SDK directly.
Rows carry `tenantId`; a query without it is a bug, not a style choice.
Services return `Result<T>`; only route handlers throw.
`replay-cache.ts` has caused three incidents; treat changes there as high risk.
Included
More in Repo memory
Learnings
Teach inline a rule from a pull request. It drafts rules from what your team dismisses and fixes, and retires the ones the evidence stops backing.
Read moreRepository wiki
A wiki generated from each repository and rebuilt on every push, plus a check that quotes the docs a merged pull request made wrong.
Read more
Related
The review
Line-anchored findings with a severity, a confidence and a validation pass behind them, from a review that reads the code around the diff.
Read moreEngineering metrics
DORA figures, cycle time split by stage, and how much of the work an AI wrote, counted from the pull requests and deploys inline already sees.
Read moreCode graph
A commit-anchored graph of symbols, calls and imports that the reviewer walks instead of grepping, following the real dependency closure of a change.
Read more
Put it on one repository
Connect GitHub or GitLab, enable a repo, and the next pull request opened gets a review.