Skip to content
NEXDGE
Start free
/01  AI pull request review

Reviewedbefore you look.

The pull request is where code review is supposed to happen and where it usually degrades into an approval. Five specialists reading it first means the human review starts from something other than a blank diff.

AI pull request review
AI pull request review is the automatic analysis of a pull request by a language model at the moment it opens, reporting security, reliability, logic, performance and quality problems in the proposed change before a human reviewer reads it. It is not an approval mechanism and does not replace a reviewer: it establishes what is mechanically wrong so the human conversation can be about design instead.
/02  Context

Why thisis a problem.

What pull request review is actually for

In most teams the pull request is where review is supposed to happen and is not really happening. A reviewer opens a diff of four hundred lines, scans it, sees nothing obviously broken, and approves. That is not negligence, it is the honest limit of attention on a Tuesday afternoon.

The result is that the mechanical pass, the one a machine is genuinely better at, never gets done, and the judgement pass gets done badly because it is competing with the mechanical one for the same attention.

A specialist review posted before a human opens the diff separates the two. What is left for the person is the part that needs a person.

One platform, said plainly

The Nexdge GitHub App is the only Git integration that exists today. There is no GitLab, Bitbucket or Azure DevOps support, and if your team is on one of those then this is not a tool you can adopt, whatever the rest of this page says.

The comparison with CodeRabbit covers that difference honestly, including the part where they integrate with all four and we integrate with one.

/03  Coverage

What the review reports

  • Findings by severity

    Each problem carries a severity from CRITICAL down to LOW, defined by exploitability and impact rather than by how alarming it reads.

  • A weighted score

    One number across the five specialists, with Security and Reliability weighted more heavily, so a pipeline can gate on it.

  • A synopsis

    Three sentences of plain English on what the change does and what is wrong with it, for the reviewer who is deciding whether to read further.

  • PASS entries

    Patterns the change got right, marked explicitly, so the report is not only ever a list of complaints.

  • A verified rewrite

    Corrected code for what failed, scanned again before it is offered.

/04  How

Setting it up

  1. Install the GitHub App

    Install Nexdge from GitHub and grant it access. Repository selection is yours: it can be a single repository rather than the whole organisation.

  2. Choose what it reviews

    Reviews run on pull requests, and a full audit can be run against any branch when you want the whole picture rather than one change.

  3. Open a pull request

    The review runs automatically and the result appears on the pull request. Credits are deducted on the same line-count scale as any other review, and only after the analysis succeeds.

  4. Gate on the result, if you want to

    The score and risk level are available through the public API, so a CI job can fail a build on them rather than treating the review as advisory.

/05  Limits

What itdoes not do.

  • GitHub only. No GitLab, no Bitbucket, no Azure DevOps, and no timeline being promised here for any of them.
  • It reviews the change, not the system. A pull request that is individually clean can still be wrong for the architecture it lands in.
  • It is not an approval. Nothing here should be wired to auto-merge, and a green review is not a reviewer.
  • Credits scale with line count, so a very large pull request costs more than a small one.
/06  Questions

The onesworth asking.

What is AI pull request review?
AI pull request review is the automatic analysis of a pull request by a language model when it opens, reporting security, reliability, logic, performance and quality problems in the change before a human reviewer reads it. It establishes what is mechanically wrong so the human review can be about design.
Which platforms does Nexdge support?
GitHub only, through the Nexdge GitHub App. There is no GitLab, Bitbucket or Azure DevOps integration today. If your team is on one of those, CodeRabbit covers all four and is the honest recommendation.
Does it replace a human reviewer?
No, and it should not be wired as though it does. It does the mechanical pass properly, which is the part human reviewers do worst when they are tired. Whether the change is the right change is a judgement call that stays with a person.
Can it block a merge?
Indirectly. The score and risk level are available through the public API, so a CI job can read them and fail a build. Nexdge itself does not enforce a merge policy.
Does the GitHub App keep our code?
No. Code from a pull request is processed in an ephemeral environment and destroyed at the end of the review, the same as any other review. Nothing is written to disk and nothing is retained in our database.
Start

Three reviews free.
No card.

Paste a real file and read what five specialists say about it. About thirty seconds.

Start for free