# Reviews that understand the code around each change

MergeHarbor indexes each repository, follows what depends on the code a pull request changes, and tries to disprove every finding before it posts a comment.

[Start free trial](https://app.mergeharbor.com/signup) [Read the docs](https://mergeharbor.com/docs/review-tiers)

mergeharbor bot An illustrative review comment

pay/refund.go:42 · major · correctness

Refund() no longer checks that the amount is at most the original charge. The guard was removed in this change.

Callers that pass customer input: api/refunds.go:118, jobs/retry.go:57

Suggested change:

\+ if amount > charge.Amount {

\+ return ErrRefundTooLarge

\+ }

## How it works

1.  ### Index the repository

    MergeHarbor builds a code graph, a text index and a semantic index of each repository. After each push, only the changed files are indexed again.

2.  ### Investigate the change

    The reviewer reads the diff, then decides what else to read: callers of a changed function, similar code, the file's history and your rules.

3.  ### Verify before posting

    A second pass tries to show that each finding is wrong, citing the code it checked. Disproved findings are never posted.

## What you get

Riskiest changes first

Each part of the diff is ranked by what it touches, such as removed guards, queries, concurrency, money and authentication, before the model reads it.

Blast radius in the summary

MergeHarbor follows changed functions, types and settings to their callers, implementations and config readers, and says how much depends on the change.

A tier for every pull request

Standard, Deep and Max use 1, 3 and 10 credits. Auto picks a tier from size, risky areas, labels and history, and says why it chose it.

The same bar at every tier

Deeper tiers investigate more, not comment more. Notes on naming, style or missing tests are held back unless you set strictness to high.

Serious findings posted early

Critical and major findings the review is highly confident about appear on the pull request before the full review finishes.

Walkthrough and questions

Larger pull requests get a step-by-step walkthrough. Ask MergeHarbor about a change in the console or with @mergeharbor on the code host.

Documentation: [Review tiers and credits](https://mergeharbor.com/docs/review-tiers) [Pull requests in the console](https://mergeharbor.com/docs/pull-requests)

## Frequently asked questions

Does a deeper tier post more comments?

No. A higher tier investigates more, but every tier posts against the same bar. Anything the Max tier's final pass adds is posted only when the check confirms it.

What happens to findings the verifier disproves?

They are never posted. The share of draft findings the second pass stopped is shown under Insights → Quality as Stopped by verifier.

Do reviews wait for the first index to finish?

No. Reviews start immediately, and indexing adds context as it completes. After the first pass, only changed files are processed.

Have another question? [Ask our support team](https://mergeharbor.com/contact?topic=support)

## Explore other features

### Merge gate

One safe-to-merge verdict with a checklist

[Merge gate](https://mergeharbor.com/features/merge-gate)

### Rules and learning

Plain-English rules and lessons from team feedback

[Rules and learning](https://mergeharbor.com/features/learning)

### Change risk

Risk scores, test gaps and migration safety

[Change risk](https://mergeharbor.com/features/change-risk)

## Review every pull request with the context of your whole codebase

Connect a code host in minutes. Reviews start right away, and codebase context is added as indexing completes.

[Start free trial](https://app.mergeharbor.com/signup) [Book a demo](https://mergeharbor.com/contact?topic=sales&source=demo)