This guide describes the highest impact changes you can make to improve the security of your code. Each section outlines a change you can make to your processes to improve security. The highest impact changes are listed first.
Key risks in the development process include:
- Using dependencies with security vulnerabilities that an attacker could exploit.
- Leaking authentication credentials or a token that an attacker could use to access your resources.
- Introducing a vulnerability to your own code that an attacker could exploit.
These risks open your resources and projects to attack and those risks are passed directly on to anyone who uses a package that you create. The following sections explain how you can protect yourself and your users from these risks.
You can secure the code you depend on by creating a vulnerability management program for dependencies. At a high level this should include processes to ensure that you:
Create an inventory of your dependencies.
Know when there is a security vulnerability in a dependency.
Assess the impact of that vulnerability on your code and decide what action to take.
As a first step, you want to make a complete inventory of your dependencies. The dependency graph for a repository shows you dependencies for supported ecosystems. If you check in your dependencies, or use other ecosystems, you will need to supplement this with data from 3rd party tools or by listing dependencies manually. For more information, see "About the dependency graph."
Dependabot can help you by monitoring your dependencies and notifying you when they contain a known vulnerability. You can even enable Dependabot to automatically raise pull requests that update the dependency to a secure version. For more information, see "About Dependabot alerts" and "About Dependabot security updates".
When you discover you are using a vulnerable dependency, for example, a library or a framework, you must assess your project's level of exposure and determine what action to take. Vulnerabilities are usually reported with a severity score to show how severe their impact could be. The severity score is a useful guide but cannot tell you the full impact of the vulnerability on your code.
To assess the impact of a vulnerability on your code, you also need to consider how you use the library and determine how much risk that actually poses to your system. Maybe the vulnerability is part of a feature that you don't use, and you can update the affected library and continue with your normal release cycle. Or maybe your code is badly exposed to risk, and you need to update the affected library and ship an updated build right away. This decision depends on how you're using the library in your system, and is a decision that only you have the knowledge to make.
Code often needs to communicate with other systems over a network, and requires secrets (like a password, or an API key) to authenticate. Your system needs access to those secrets to run, but it's best practice to not include them in your source code. This is especially important for repositories to which many people might have access and critical for public repositories.
Note: Secret scanning is available for organization-owned repositories in GitHub Enterprise Server if your enterprise has a license for GitHub Advanced Security. 詳細については、「GitHub Advanced Security について」を参照してください。
注: この機能を使用するには、サイト管理者が your GitHub Enterprise Server instance の secret scanning を有効にする必要があります。 詳しくは、「アプライアンスでの secret scanning の構成」をご覧ください。
You can configure secret scanning to check for secrets issued by many service providers and to notify you when any are detected. You can also define custom patterns to detect additional secrets at the repository, organization, or enterprise level. For more information, see "About secret scanning" and "Secret scanning patterns."
Besides your code, you probably need to use secrets in other places. For example, to allow GitHub Actions workflows or Dependabot to communicate with other systems. For more information on how to securely store and use secrets, see "Encrypted secrets in Actions" and "Managing encrypted secrets for Dependabot."
Note: Code scanning is available for organization-owned repositories in GitHub Enterprise Server. This feature requires a license for GitHub Advanced Security. 詳細については、「GitHub Advanced Security について」を参照してください。
注: この機能を使用するには、サイト管理者が your GitHub Enterprise Server instance の code scanning を有効にする必要があります。 詳しくは、「アプライアンスでの code scanning の構成」をご覧ください。
You can improve the quality and security of your code by ensuring that all pull requests are reviewed and tested before they are merged. GitHub has many features you can use to control the review and merge process. To get started, see "About protected branches."
Insecure code patterns are often difficult for reviewers to spot unaided. In addition to scanning your code for secrets, you can check it for patterns that are associated with security vulnerabilities. For example, a function that isn't memory-safe, or failing to escaping user input that could lead to an injection vulnerability. GitHub offers several different ways to approach both how and when you scan your code. To get started, see "About code scanning."