Skip to content

Claude Code Review adds Quick, Standard and Deep review levels

Version 2.1.304 lets organisations set a default review depth for Code Review and override it per repository, plus a new option to cover every repo in a GitHub org automatically.

Source: Anthropic

Claude Code 2.1.304, shipped 19 October, gives organisation owners a choice of how thoroughly Code Review checks a pull request: Quick, Standard, or Deep. Standard stays the default, and owners can set the organisation-wide level while giving individual repositories their own.

The same release adds an admin setting that covers every repository in a connected GitHub organisation automatically, including ones created later, so nobody has to remember to add a new repo to the review list by hand. A smaller fix in the same build stops pull requests opened within a minute or two of an admin turning Code Review on from being silently skipped.

Code Review level: Quick | Standard (default) | Deep
Set per organisation, overridable per repository

Why it matters: teams running Code Review across many repositories can now trade off depth against turnaround time where it matters, rather than applying one setting everywhere, and new repositories get covered without manual setup.

The caveat: Anthropic’s changelog names the three levels but doesn’t document what actually differs between them, so you can’t yet tell from the release notes whether Deep means more review time, higher token cost, or wider scope.

Share

Related reading

ML for Web Devs

Machine learning for web developers: the only mental model you need

You do not need calculus to work with models. You need one idea, that a model is a function you fit instead of write, and the rest follows from things you already know.

ML for Web Devs

What an embedding actually is

An embedding is a list of numbers that puts similar things near each other. That one sentence is enough to build search, recommendations and retrieval. Here is how.

ML for Web Devs

Tokens, context windows, and why your bill is what it is

The unit of billing for language models is not the request or the word. Once you understand tokens and how a context window fills up, the invoice stops being a surprise.

More from The Wire

All briefs

GitHub's new Copilot default policy takes effect today

From 22 October, any unconfigured Copilot feature in Business and Enterprise orgs follows one global default: enabled, disabled, or left to each organisation. Admins had a 28-day window to set it before this landed.

via GitHub