> For the complete documentation index, see [llms.txt](https://docs.nullify.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.nullify.ai/capabilities/code-reviews.md).

# Code Reviews

## Overview

Nullify reviews code across three layers: application-code vulnerabilities, dependency vulnerabilities, and leaked secrets. Application-code and dependency review run on pull requests and on the default branch, while secrets are checked on every commit across every branch — keeping security issues from reaching production.

<figure><picture><source srcset="https://3946391995-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FId2OtLCL6O6V3XSRYxtr%2Fuploads%2Fgit-blob-3ca86832f15f6ebd07649cbf9cbf1eced1cbcfad%2Fcode-review-detail-dark.png?alt=media" media="(prefers-color-scheme: dark)"><img src="https://3946391995-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FId2OtLCL6O6V3XSRYxtr%2Fuploads%2Fgit-blob-e672a4d67b6e8a5d5ecc466248c782ae40141a27%2Fcode-review-detail.png?alt=media" alt="Code review checks and open findings"></picture><figcaption><p>Each scanner's check and the repository's open findings.</p></figcaption></figure>

## Scanning Scope

### Application Code Analysis

* **Pull Requests**: Reviews PRs targeting the default branch
  * Reviews the changed code in the diff
  * Posts inline review comments on findings
  * Reports a **Nullify** check whose pass/fail is [configurable](https://docs.nullify.ai/configuration/configuration-pull-requests.md) — informational by default, or a required gate on net-new findings
* **Default Branch Commits**: Reviews all commits pushed to the default branch
  * Full repository review on every push
  * Updates the dashboard with the complete inventory
  * Creates tickets and notifications for new findings
* **Other Branches**: Not reviewed for code vulnerabilities

### Dependency Analysis

* **Pull Requests**: Reviews dependency changes in PRs targeting the default branch
  * Reviews changed manifest and lock files (`package.json`, `pom.xml`, `Cargo.toml`, etc.)
  * Contributes to the same tenant-configurable **Nullify** check gate
  * Identifies both direct and transitive dependencies
* **Default Branch Commits**: Full dependency review on every push
  * Analyzes all package manifest and lock files
  * Updates the dashboard with the complete vulnerability inventory
  * Drafts autofix PRs for upgradeable dependencies
* **Container Images**: Reviews container images for vulnerable packages

### Secrets Detection

* **Pushes and Pull Requests**: A standalone scan inspects the event's commits
  * Diff-scoped review
  * Notifies the commit author when secrets are found
  * Actively verifies detected credentials
  * Reports newly committed secrets so they can be revoked and removed
* **Repository Inventory**: An initial full-history scan establishes the inventory. After a full-history scan has completed, later inventory scans can use shallow HEAD snapshots
  * Updates the dashboard with a repository-wide secret inventory

## Code reviews in the dashboard

Open **Assessments → Code reviews**. The workspace has these tabs:

| Tab                                                                                    | What it shows                                                                                                                                                                                                                    |
| -------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Runs**                                                                               | Each code review of a pull request or commit, with its status and finding count. Filter by repository, branch, commit, pull request number, status and start date. Open a review to see its findings and its **Merge controls**. |
| **Pipelines**                                                                          | One row per pipeline run for a repository, branch, commit and trigger. Each row shows its worst Nullify check and its new, fixed and reintroduced findings. Expand a row to see each scanner that ran.                           |
| **Pull request blockers**                                                              | Pull requests that a Nullify check or change request blocked in the last 14 days, and those whose review status Nullify could not verify. Each row shows the reason, the gate mode and the live check.                           |
| **Code findings**, **Credentials**, **Dependencies**, **Containers**, **Supply chain** | One tab per scanner. Each lists repositories with their open findings by severity, last commit and last assessment. Expand a repository for its recent scan history, or open its findings.                                       |
| **Check runs**                                                                         | The checks Nullify published, filtered by time window (**Last day** by default), outcome, repository and service.                                                                                                                |

Admins can mark a blocking Nullify check as passing from **Merge controls**, **Pull request blockers**, **Pipelines** or **Check runs**, and withdraw Nullify's change requests from **Merge controls**. See [Merge overrides](https://docs.nullify.ai/configuration/configuration-pull-requests.md#merge-overrides).

### Run a review on demand

Select **Run code assessment**. It needs the **Editor** or **Admin** role. Choose a **Repository** and an **Assessment** type:

* **Review an open pull request** reviews the current head commit of the pull request whose number you enter. The findings and merge controls appear on the new review.
* **Default-branch scanner assessment** scans the head of the default branch with the **Scanners** you select: **Code**, **Dependencies and containers**, **Credentials** and **Supply chain**, all selected by default. The selected scanners use your tenant's triage and budget settings. The run also updates the commit's check, the repository context and the SBOM.

Select **Start assessment** to open the new run. If a run is already in progress for that branch or pull request, Nullify opens that run instead of starting a second one. On a finished review, **Re-run pull request review** or **Run another code assessment** reopens the dialog with the repository and pull request filled in. The API equivalents are [Start a repository scan](https://docs.nullify.ai/api-reference/api-reference/orchestrator.md#start-a-repository-scan) and [Start a pull request review](https://docs.nullify.ai/api-reference/api-reference/orchestrator.md#start-a-pull-request-review).

### Feedback on review comments

Nullify's review comments on code, dependency, container and secret findings end with **Was this useful?** and thumbs-up and thumbs-down links. The descriptions of Nullify's code, dependency and container fix pull requests carry the same links. Anyone who can read the comment can use them, without a Nullify login.

The link opens a feedback page that records the vote when it loads. The page shows nothing about the finding. You can then pick reasons, add a note of up to 2,000 characters and select **Send**:

* After a thumbs-up: **Useful finding**, **Worth fixing here**, **Fix looks right**, **Clearly explained** and **Severity looks right**.
* After a thumbs-down: **Not a real issue**, **Not worth raising here**, **Fix is wrong**, **Hard to follow** and **Severity is off**.

**Actually, it wasn't useful** (or **Actually, it was useful**) records the opposite vote and shows the other set of reasons. A link that is no longer valid shows **That link has expired**.

Feedback never changes the finding, its status or its priority. The **Feedback** section of the finding's details counts the responses and notes those from pull request links. See the [Finding Feedback API](https://docs.nullify.ai/api-reference/api-reference/finding-feedback.md) to read them.

## What Nullify covers

* **Languages & frameworks** — a wide range of languages plus infrastructure-as-code. See [Supported Languages](https://docs.nullify.ai/capabilities/code-reviews/code-analysis/supported-languages.md).
* **Dependency ecosystems** — npm, Maven, Go, PyPI, Cargo, and more, with reachability on supported languages. See [Supported Ecosystems](https://docs.nullify.ai/capabilities/code-reviews/dependency-analysis/supported-ecosystems.md).
* **Weakness classes** — OWASP Top 10 and CWE Top 25 vulnerability classes, plus compound attack paths. See [Supported Weaknesses](https://docs.nullify.ai/capabilities/code-reviews/code-analysis/supported-weaknesses.md).

## Reachability & Exploitability Analysis

Nullify doesn't just report what a scanner matched — it reasons about whether a finding actually matters.

### Code Analysis

* AI-driven triage analyzes how code is used in context.
* Screens out false positives before they reach you.
* Severity reflects exploitability, not just theoretical risk.

### Dependency Analysis

* **Direct vs. Transitive**: marks whether the vulnerability is in a direct dependency.
* **Reachability Analysis**: on supported languages, traces whether a vulnerable package is actually reached by your code.
* Transitive and unreachable vulnerabilities are prioritized lower by default.

### Secrets Verification

* **Active validation**: checks discovered credentials against the real provider.
* Tests API keys against actual services and validates token formats and expiration.
* Not just pattern matching — functional testing. See [Secrets Detection](https://docs.nullify.ai/capabilities/code-reviews/secrets-detection.md).

Findings from all three layers are [triaged](https://docs.nullify.ai/capabilities/triage.md) together, then routed and remediated through [Campaigns and Strategy](https://docs.nullify.ai/capabilities/program-management.md) and [Remediations](https://docs.nullify.ai/capabilities/remediations.md).
