How the dependency graph recognizes dependencies

The dependency graph can identify your project's dependencies using the following methods.

Method	How it works
Static analysis	Parses manifest and lock files in your repository
Dependabot graph jobs	Uses a Dependabot GitHub Actions workflow to generate dependency snapshots
Automatic submission	Runs a built-in GitHub Actions workflow to resolve build-time dependencies
Dependency submission API	Accepts dependency data you submit programmatically

Once dependencies are in the graph, you can receive Dependabot alerts and Dependabot security updates for any known vulnerabilities.

Static analysis

When you enable the dependency graph, GitHub scans your repository for supported manifest files and parses each package's name and version. The graph updates when you change a supported manifest or lock file on your default branch, or when a dependency changes in its own repository.

Static analysis can identify:

• Direct dependencies explicitly defined in a manifest or lock file
• Indirect dependencies—dependencies of these direct dependencies, also called "transitive dependencies"—but only if they are defined in a manifest or lock file, not if they are resolved at build time

For the most reliable graph, you should use lock files (or their equivalent), because they define exactly which versions of the direct and indirect dependencies you currently use. Lock files also ensure that all contributors to the repository are using the same versions, which will make it easier for you to test and debug code. In addition, indirect dependencies inferred from manifest files (rather than lock files) are excluded from vulnerability checks.

Automatic dependency submission

Some ecosystems resolve indirect dependencies at build time, so static analysis can't see the full dependency tree. When you enable automatic dependency submission for a repository, GitHub automatically identifies the transitive dependencies in the repository for supported ecosystems. See Dependency graph supported package ecosystems.

In the background, automatic dependency submission runs a GitHub Actions workflow that generates the complete tree and uploads it using the dependency submission API. Automatic dependency submission runs on GitHub-hosted runners by default and counts toward your GitHub Actions minutes. Optionally, you can choose to run it on self-hosted runners or larger runners.

To enable automatic dependency submission, see Configuring automatic dependency submission for your repository.

Dependabot graph jobs

This method uses a special type of Dependabot job that builds a dependency snapshot and uploads it to the dependency submission API. This is currently only supported for Go dependencies.

This approach is similar to automatic dependency submission, but does not incur charges for GitHub Actions minutes. It can also access organization-wide configurations for private registries you've set up for Dependabot.

The dependency submission API

You can call the dependency submission API in your own script or workflow. This is useful if:

• You need to submit transitive dependencies that cannot be detected from lock files.
• You need to create custom logic or are using an external CI/CD system.

Dependencies are submitted to the dependency submission API in the form of a snapshot. This is a list of dependencies associated with a commit SHA and other metadata, reflecting the current state of your repository.

If you are calling the API in a GitHub Actions workflow, you can use a pre-made action for your ecosystem that automatically gathers the dependencies and submits them to the API. Otherwise, you can write your own action or call the API from an external system.

Submitted dependencies will be shown in dependency review, but are not available in your organization's dependency insights.

For more information, see Using the dependency submission API.

Prioritization

A repository can use multiple methods for dependency submission, which can cause the same package manifest to be scanned multiple times, potentially with different outputs from each scan. Dependency graph uses deduplication logic to parse the outputs, prioritizing the most accurate information for each manifest file.

Dependency graph displays only one instance of each manifest file using the following precedence rules.

1. User submissions take the highest priority, because they are usually created during artifact builds they have the most complete information.
   • If there are multiple manual snapshots from different detectors, they are sorted alphabetically by correlator and the first one used.
   • If there are two correlators with the same detector, the resolved dependencies are merged. For more information about correlators and detectors, see REST API endpoints for dependency submission.
2. Automatic submissions have the second-highest priority since they are also created during artifact builds, but are not submitted by users.
3. Static analysis results are used when no other data is available.