Skip to content

Docs / Migrating from CodeRabbit

Migrating from CodeRabbit

Move review onto your own key without rewriting config — your .coderabbit.yaml keeps working.

You do not have to rewrite your configuration, and you do not have to switch in one step. The reviewer reads your existing .coderabbit.yaml, which means you can run both, compare, and only then turn one off.

If you want the commercial argument rather than the procedure, that is on developerz.ai vs CodeRabbit.

Run both first

You do not need to uninstall anything to start. developerz.ai recognises CodeRabbit by sender identity and defers to it on any pull request where it is active — no bot-on-bot loops, and no second bot arguing with the first on the same thread.

That is what makes an evidence-based migration possible: install our App alongside, watch a few real pull requests, and compare what each reviewer flags before you commit to anything.

Then switch

  1. Uninstall the CodeRabbit GitHub App from the repositories you are moving.
  2. Keep .coderabbit.yaml exactly as it is. Nothing needs to change.

That is the whole migration. The developer experience stays the same; the inference moves onto your own provider account.

What carries over

Most of the config is honoured directly — language, tone_instructions, the reviews.profile severity floor, path_filters, path_instructions, the whole auto_review block (incremental review, draft handling, base branches, ignored titles and logins, label gates), chat.auto_reply, and the knowledge_base scopes.

Some keys are recognised but not yet acted on — code generation, issue enrichment, custom summary and labelling instructions, reviewer suggestion and auto-assignment, mutually-exclusive label groups. When one of those is present the reviewer posts a warning comment rather than silently ignoring it, so the gap stays visible instead of becoming a surprise three weeks later.

Anything not recognised at all surfaces an unknown key warning.

The full, generated field table lives on the AI code review page and is regenerated from the code, so it cannot drift from what actually runs.

Optional: move to the native config

Once you are settled you can delete .coderabbit.yaml and use the review: block in .maintainer.yml instead, which puts review policy in the same file as everything else the agent does. There is no rush and no functional penalty for staying on the CodeRabbit file.

What changes about the bill

Review stops being a per-user seat plus a per-file charge and becomes part of the flat per-AI-dev price. The variable cost that remains is your own token spend, against your own key, invoiced by your own provider. A 400-file pull request costs the same to review as a one-line one.

See billing and BYOK.

What does not change

  • Review comments still land on the pull request, in the same place.
  • Every comment discloses that it is automated. That is not configurable.
  • Your branch protections stay the gate. Review can block a merge on paid tiers, but it cannot merge past a protection you set.