GitHub

Cortex connects to many third-party vendors whose system interfaces frequently change. As a result, integration behavior or configuration steps may shift without notice. If you encounter unexpected issues, check with your system administrator or refer to the vendor's documentation for the most current information. Additionally, integration sync times vary and are subject to scheduling overrides and timing variance.

Why use the integration for GitHub

GitHub is a Git-based source code management platform that helps teams share code, track changes over time, and collaborate on projects across an organization.

Integrating GitHub with Cortex allows you to:

• Automatically discover GitHub repositories and track ownership of the entities they represent in Cortex

• Manage your Cortex workspace through a GitOps workflow, using GitHub as the source of truth for entity definitions

• View repository context directly on an entity's details page, including the associated repo, recent commits and releases in the event timeline, the most-used language, and top contributors with their contribution counts

  • In an entity's Code & Security section, surface vulnerabilities from GitHub Advanced Security, including code scanning, Dependabot alerts, CodeQL, and secret scanning

• Track pull requests and work items from the engineering homepage to keep delivery work visible alongside the rest of your catalog

• Use GitHub metrics in Eng Intelligence to understand service health, incident response patterns, and other key engineering signals

• Create Scorecards that drive alignment and measure progress on initiatives involving your repositories

You can also optionally integrate GitHub Copilot with Cortex to:

• Track Copilot adoption and measure its impact on engineering productivity within Eng Intelligence

GitHub Copilot integration

If you installed the GitHub app prior to October 14, 2025, you must accept new permissions in order to access Copilot metrics in Cortex.

In your email, locate the message with subject line "[GitHub] Cortex App is requesting updated permissions" from [e-mail] and click the link to review and accept the change. As a security best practice, always verify that the email was sent from a trusted domain. In GitHub, click Accept new permissions.

Alternately, you can remove your GitHub integration in Cortex and re-configure it. The GitHub app is preconfigured to include the permissions necessary for the Copilot integration.

Configuring GitHub

Prerequisites

If you connect Cortex via a custom GitHub app, it must be configured with a fine-grained personal access token containing the minimum permissions listed below. Note that the Cortex GitHub app is preconfigured with these permissions.

Repository permissions

Permission	Requirement	Purpose in Cortex
Actions	Read & write	Read workflow run information for Git-based CQL rules Artifact information for actions
Administration	Read & write	Create repositories
Checks	Read & write	Create and update check runs for linting and validation workflows, such as the cortex.yaml linter on pull requests
Code scanning alerts	Read-only	Get vulnerability information for Git-based CQL rules
Commit statuses	Read & write	Read commits for an entity's Git metadata Read commits for Git-based CQL rules Show pending status messages on the OpenAPI incompatibility check
Contents	Read & write	Read cortex.yaml, cortex-properties.yaml, and package/OpenAPI files Read Git rules Create file contents
Custom properties	Read & write	Used in Workflows
Dependabot alerts	Read-only	Read vulnerability information for Git CQL rules (only relevant if using Dependabot)
Deployments	Read & write	Used in Workflows
Issues	Read & write	Read associated issues with repositories for populating entity Git integration and for Git CQL rules Create new issues based on Initiatives
Metadata	Read-only	Read associated data with repositories for populating entity Git integration and for Git CQL rules
Pull requests	Read & write	Read pull request data for Git CQL rules and developer homepage "My PRs" tab Comment if there are breaking OpenAPI changes on a PR
Secret scanning alerts	Read-only	Read vulnerability information for secret scanning
Secrets	Read & write	Optionally write repo secrets after creating new repo
Single file	Read & write (path to cortex.yaml)	Read cortex.yaml files Create cortex.yaml files
Workflows	Read & write	Write in GitHub actions files

Organization-level permissions

Permission	Requirement	Purpose in Cortex
Administration	Read-only	Create repositories Pull in Copilot data for Eng Intelligence
Members	Read & write	Read membership information for ownership and team composition Write permission used in Workflows
GitHub copilot business	Read-only	Read copilot-related metrics for AI Impact Dashboard

Enabling PR webhooks (optional but recommended)

Cortex ingests pull request (PR) data via the GitHub API. Subscribing to GitHub's PR webhooks adds a real-time stream of PR events that Cortex cross-references against API data, which is the most efficient way to ensure every PR and event is captured without gaps.

If you use the Cortex GitHub app, no action is required. The Cortex GitHub app receives these PR webhook events by default, so you get real-time PR data out of the box.

If you use a custom GitHub app or a personal access token (PAT), you must subscribe to the following events manually at the repository or organization level:

• Pull requests

• Pull request reviews

• Pull request review comments

• Pull request review threads

• Issue comments (GitHub delivers PR comments as issue-comment events)

• Organizations

• Repositories

For a custom GitHub app, select the above events under your app's Permissions & events > Subscribe to events, then click Save.

For a PAT connection, add the above events to your Cortex webhook under the repository or organization's Settings > Webhooks.

Cortex begins receiving PR webhook events immediately and automatically cross-references them against standard API ingestion to validate the data.

Choosing a configuration option

There are multiple options for connecting Cortex to your GitHub instance:

• Through Cortex's official GitHub app

  • This option is supported for use with a single organization in GitHub.

• Through a custom GitHub app

  • By using Cortex's official GitHub app or a custom GitHub app, users can tag entities with Git details and enable GitOps-style configuration of data in Cortex.

• Using a personal access token

• Using Cortex Axon Relay, a relay broker that allows you to securely connect your on-premises GitHub data.

If your GitHub setup involves multiple organizations, you can add multiple GitHub apps, use a personal access token that has access to all orgs, or create multiple configurations with corresponding aliases.

See the tabs below for instructions on each of the GitHub integration options.

If you're using a self-hosted instance of GitHub, you'll need to verify that your Cortex instance is able to reach the GitHub instance. Requests are routed through a static IP address.

Contact support at [e-mail] to receive details about our static IP. If you're unable to directly allowlist our static IP, you can route requests through a secondary proxy in your network that has this IP allowlisted and have that proxy route traffic to your GitHub instance.

Cortex GitHub app

Configure GitHub with the Cortex GitHub app

Cortex's official GitHub app is the easiest way to connect your GitHub instance. It is preconfigured with the permissions needed to use this integration. Note that it is not available for Cortex Server.

To set up the app:

1. In Cortex, navigate to the GitHub settings page:

   1. Click Integrations from the main nav. Search for and select GitHub.

2. On the GitHub settings page, next to cortex, click Install.

3. Follow the prompts, then click Install & Authorize.

Cortex's GitHub app is preconfigured with:

• Permissions for catalogs, Scorecards, and the Scaffolder.

• Webhooks to enable GitOps.

• Support for using GitHub teams as an ownership provider.

The app comes with a built-in linter, which validates the format of a given cortex.yaml file and checks Cortex-specific items. However, the linter DOES NOT validate data correctness. For example, the linter will confirm that the format of a group block is correct, but will not check that the group exists.

Custom app

Configure GitHub with a custom app

If you're using Cortex Server, or if you don't want to use the official Cortex app, you can connect a custom GitHub app.

Make sure you have configured the permissions listed above under Prerequisites.

Step 1: Register the custom app in GitHub

1. Register the app. When you're creating the app, make sure to follow these steps:

   • Disable "Expire user authorization tokens." (Cortex does not support this OAuth workflow yet.)

   • Select "Request user authorization (OAuth) during installation."

   • Callback URL: https://app.getcortexapp.com/github/redirect/{alias}

     • Make sure to use the GitHub configuration alias and not the tenant name in the URL.

   • Webhook URL: https://api.getcortexapp.com/api/internal/v1/github/webhook

   • Add a webhook secret of your choosing that will be identical to the webhook secret token configured later in Cortex settings.

   • Set the repository and organization permissions outlined in the configuration modal.

   • Check Push and Check suite under Subscribe to events.

2. After the app has been created, generate a client secret and private key.

Step 2: Configure the custom app in Cortex

1. In Cortex, navigate to the GitHub settings page:

   • Click Integrations from the main nav. Search for and select GitHub.

2. Click Add configuration, then for the type, select GitHub app.

3. Fill in the form:

   • Alias: Enter the name Cortex will associate with a given configuration.

   • Application ID: Enter the ID for your custom GitHub ap.

   • Client ID: Enter the unique client ID assigned to your app during registration.

   • Client secret: Enter the unique client secret assigned to your app during registration.

   • Private key: Enter the private key to authenticate with your GitHub app.

   • Public link: Enter the public URL of your GitHub app.

   • API endpoint (for GitHub enterprise): Enter the endpoint root URL for self-managed GitHub setups.

4. Click Save.

5. On the GitHub Settings page in Cortex, under Webhook, choose your new configuration alias and enter the same Secret token that was entered in GitHub during the app configuration.

6. On the GitHub Settings page in Cortex, next to your custom app configuration's name, click Install.

Personal access token

Configure GitHub with a personal access token

Prerequisites

Before getting started, make sure your personal access token has, at minimum, the repo and read:org permissions.

Note: Beyond the minimum permissions indicated above, many scenarios will require that the user who generated the personal access token have organization ownership permissions within GitHub.

Configuration

1. In Cortex, navigate to the GitHub settings page:

   • Click Integrations from the main nav. Search for and select GitHub.

2. Click Add configuration.

3. In the upper right corner of the modal, click the dropdown and select Personal access token.

   
4. Fill in the form:

   • Alias: Enter a name that Cortex will associate with a given configuration.

   • Token: Enter the Personal access token generated in GitHub.

   • API endpoint (for GitHub enterprise): Enter the endpoint root URL for self-managed GitHub setups.

5. Click Save.

Relay broker

Configure GitHub with Cortex Axon Relay

See Internally hosted integrations for instructions. Make sure to follow the GitHub-specific instructions for the docker-compose.yml file.

Configuring the integration for multiple GitHub accounts​

The GitHub integration has multi-account support. You can add a configuration for each additional organization, instance, or account by repeating the process above.

Each configuration requires an alias, which Cortex uses to correlate the designated organization, instance, or account with registrations for various entities. Registrations can also use a default configuration without a listed alias. You can edit aliases and default configurations from the GitHub page in your Cortex settings.

To set the default configuration:

1. From the main sidebar, select Integrations.

2. Locate GitHub, then click Settings.

3. Locate the configuration, then click the pencil icon.

4. From the Category drop-down menu, select a category. You must select at least one category before you can set the configuration as default.

5. Toggle on Set as default. > Tip: If you only have one configuration, it's automatically set as the default.

Cortex supports mapping multiple identities for a single user if you have multiple configurations of GitHub. See Configuring identity mappings for more information.

To write rules related to Dependabot alerts, you must verify the necessary permissions are set for repositories you'd like to see vulnerabilities reported on.

To verify, navigate to a GitHub repo and click Settings > Code security and analysis. Ensure you are a member of a team under Access to alerts.

Registration

See the Create services documentation for instructions on importing entities.

Entity descriptor

Repository

You can define a GitHub repository for a given entity by adding the x-cortex-git block to the entity's descriptor. When you define a repository, Cortex checks for Security Advisory vulnerabilities from the GraphQL API and Advanced Security vulnerabilities from the Rest API.

x-cortex-git:
  github:
    repository: cortex/docs
    basepath: myService
    alias: myApp

Field	Description	Required
repository	GitHub repository in the form /	✓
basepath	Subdirectory for the entity if it is in a monorepo. Note that setting a basepath filters the vulnerabilities that appear in Cortex; Advanced Security vulnerabilities will not appear.
alias	Alias for the configuration in Cortex (only needed if you have opted into multi-account support)

Only one repository can be defined for in a given entity's YAML in the x-cortex-git block. Users looking to list additional repositories without the full functionality of GitOps can define the repos as custom data.

x-cortex-custom-metadata:
  second-git-repo:
    - `/`
  third-git-repo:
    - `/`

Ownership

You can define the following block in your Cortex entity descriptor to add your GitHub teams. Be sure to include both your GitHub organization name and the team name in the name field.

x-cortex-owners:
  - type: group
    name: cortex/engineering
    provider: GITHUB
    description: This is a description for this GitHub team that owns this entity.

Field	Description	Required
type	Ownership type; must be defined as group for GitHub teams	✓
name	GitHub team name in the form /Team names are generally converted to lowercase with - separators (Team Name would be cortex/team-name), but you can verify your exact name from the GitHub permalink	✓
provider	Name of integration (in this case, GITHUB)	✓
description	Description for the GitHub team

Multiple GitHub organizations are not supported for ownership, and Cortex will use the default configuration when fetching teams.

Identity mappings

Cortex maps users' email addresses to discovered GitHub accounts, so you never need to define email ownership in an entity descriptor. Users must be members of GitHub teams to be pulled in to Cortex.

You can confirm users' GitHub accounts are connected from GitHub identity mappings in settings.

Using the GitHub integration

View GitHub data on entity pages in Cortex

The GitHub integration will populate the Repo and Language detail blocks on an entity's details page. If a GitHub team has been defined as the owner for an entity, it will also appear in the Owners block.

Code & security

Vulnerabilities appear in the Vulnerabilities block under Code & security on an entity page overview.

Click Code & security in an entity's sidebar to view the full list of vulnerabilities for an entity. Cortex checks for:

• Security Advisory vulnerabilities from the GraphQL API

• GitHub Advanced Security vulnerabilities

  • Cortex pulls data from code scanning, Dependabot alerts, CodeQL, and Secret scanning.

  • Dependency reviews are not supported.

GitHub Advanced Security vulnerabilities are not surfaced for monorepos.

You can query for vulnerabilities with CQL and create Scorecard rules based on security metrics. See Scorecards and CQL below.

Events

Recent commits appear at the top of an entity's overview page.

You can also click Events in the entity's sidebar to see all commits and releases associated with that entity. Each is hyperlinked to the commit or release page in GitHub and includes a timestamp.

Repository

You can access more detailed information pulled from GitHub in the Repository page in the sidebar. At the top of the page, you'll find the repo(s) associated with that entity and the most-used language in files for that entity. In the Top contributors block, you'll find the three users who have contributed the most code and the number of their contributions.

In the Commits section, you'll find the 10 most recent commits and metadata about each. Below Commits is the Recent releases section, which includes the 5 most recent releases.

Issue tracking

From the Issue tracking page in the entity's sidebar, you can find a list of open GitHub issues. Each issue will show the number, title, assignees, and date created.

Packages

Packages are automatically scraped from your Git repos or they can be submitted via the packages API. The package file must be in the root of your repository — or, if you're using basepath, in the root of the subdirectory — to be scraped by Cortex. You can query an entity's packages in CQL explorer using packages().

To view packages, click Packages in the entity's sidebar.

The following package types are automatically scraped from repositories:

• JavaScript / Node.js: package.json, package-lock.json, yarn.lock, pnpm-lock.yaml

• Python: requirements.txt, pipfile.lock

• .NET (C#): packages.lock.json

• Java: pom.xml

• Go: go.sum

All other files of these types can be added via the packages API.

CI/CD - GitHub workflows

From the CI/CD > GitHub workflows page in the entity's sidebar, you can find a history of GitHub workflow runs for the past week. Each run is tagged with its status: IN_PROGRESS, COMPLETED, SUCCESS, CANCELLED, FAILURE, PAUSED.

The GitHub workflows page displays data about workflows in GitHub, not Workflows initiated via Cortex's Workflows tool.

Team entity pages

When a GitHub team is registered with a team entity, Cortex will pull GitHub users in to the Members tab. When available, Cortex will pull in the profile picture and email address for each user.

Engineering homepage

The GitHub integration enables Cortex to pull information about pull requests and issues into the homepage. You can find your open pull requests, any pull requests assigned to you for review, and any issues assigned to you.

Pull requests and issues from GitHub are refreshed every 2 minutes.

Eng Intelligence

The Eng Intelligence tool uses pull request data from GitHub to generate metrics:

• Average PR open to close time

• Avg time to first review

• Avg time to approval

• PRs opened

• Weekly PRs merged

• Avg PRs reviewed/week

• Avg commits per PR

• Ave lines of code changed per PR

Eng Intelligence also pulls in data from Copilot to show AI impact in the Copilot Dashboard.

You can read more about how Eng Intelligence tracks metrics for teams and users in the Eng Intelligence documentation.

To add deployments for your Github related entity, you can send a deployment event to the Cortex API.

Bot-authored pull requests and reviews

Some pull requests and reviews in GitHub come from bot accounts, like Dependabot or a GitHub App that a team set up. Cortex ingests this activity along with everything else and attributes it to the bot that produced it, so you can include or exclude it deliberately.

How Cortex identifies a bot

Cortex uses the account type that GitHub reports. If GitHub identifies the account as a bot, Cortex marks its pull requests and reviews as bot generated. Automation that runs under a regular GitHub user account isn't marked as bot generated, because GitHub reports that account as a user.

Where bot accounts appear

• A pull request opened by a bot shows that bot's account name as the author, for example dependabot[bot], rather than an empty author.

• In Eng Intelligence, bot accounts appear alongside people in the User filter and group-by options, so you can single out or exclude a specific bot.

• Eng Intelligence metric values don't change. These pull requests were always counted, so all that's new is the author attribution.

Filtering bot activity in CQL

git.pullRequests() and git.reviews() return bot activity by default. GitPullRequest and GitReview each have an isBotGenerated field, which is true when GitHub reports the account as a bot. For more information about writing queries, refer to Cortex Query Language (CQL).

To count only the pull requests opened by people:

git.pullRequests(lookback=duration("P14D")).filter((pr) => !pr.isBotGenerated).length > 0

To look at automated review activity on its own:

git.reviews(lookback=duration("P7D")).filter((review) => review.isBotGenerated).length > 0

Availability

Cortex began marking bot activity on August 18, 2026. Pull requests and reviews that Cortex ingested before then have isBotGenerated set to false, even when a bot produced them, so a query with a long lookback period can undercount bot activity.

Scorecards and CQL

With the GitHub integration, you can create Scorecard rules and write CQL queries based on GitHub data.

See more examples in the CQL Explorer in Cortex.

Approvals required to merge

Number of approvals required to merge a pull/merge request into a repository. Defaults to 0 if no approvals are defined.

Definition: git.numOfRequiredApprovals()

Examples

For a security or development maturity Scorecard, you can write a rule to make sure at least one approval is required to merge a pull/merge request:

git.numOfRequiredApprovals() > 0

By having a rigorous PR process in place for a repo, you can make sure changes aren't made that create vulnerabilities. This kind of rule could also be used in a best practices or project standards Scorecard.

You can also use a similar expression in the Query Builder to find entities lacking approval:

git.numOfRequiredApprovals() < 1

Git repository set

Check if an entity has a registered Git repository.

Definition: git (==/!=) null: Boolean

Example

In a Scorecard, you can write a rule that detects whether an entity has a Git repository set:

git != null

Branches

List all live branches with some basic metadata.

• Head

• Is protected

• Name

Definition: git.branches()

Example

For a best practices Scorecard, you can make sure that branches associated with an entity match a standard naming convention:

git.branches().all((branch) => branch.name.matches("(main|master|feat-.*|bug-.*|task-.*)"))

Branch protection details

git.branchProtection() now returns the effective branch protection for a branch by merging policies from both classic branch protection and GitHub rulesets into a unified view. This means it reflects what GitHub actually enforces, regardless of whether a repo uses classic rules, rulesets, or a combination of both.

Find details for specified branch, or default branch if none is specified.

• Branch name

• Code owner reviews required

• Dismiss stale reviews

• Required status checks

• Restrictions apply to admin

• Review required

Definition: git.branchProtection()

Examples

For a security Scorecard, you can write a rule to make sure the default branch is protected:

git.branchProtection() != null

Because vulnerabilities in the default branch are critical, this rule should be in one of the first couple levels.

You can also use the Query Builder to find entities with unprotected default branches:

git.branchProtection() = null

Branch rulesets

GitHub branch rulesets are a newer, more flexible alternative to classic branch protection. Rulesets are layered (organization + repository), composed of typed rules, and can target multiple branches by pattern. Cortex exposes them in CQL through git.branchRulesets(...).

git.branchRulesets(branchName?) returns the effective ruleset for a branch, i.e. the rules from every active ruleset that targets that branch, merged together. Pass a branch name to target a specific branch; omit it to use the repo's default branch. Returns null if no ruleset applies.

git.branchRulesets().rules.any((rule) =>
    rule.type == "pull_request" && rule.requiredApprovingReviewCount >= 2
)

// Require Copilot code review on push for the default branch
git.branchRulesets().rules.any((rule) =>
    rule.type == "copilot_code_review" && rule.reviewOnPush == true
)

// Forbid pushes that match a banned commit-message pattern
git.branchRulesets().rules.any((rule) =>
    rule.type == "commit_message_pattern" && rule.patternOperator == "regex"
)

GitBranchRulesets fields

Field	Type	Description
branchName	String	Branch the rules apply to.
rules	List<GitBranchRule>	All effective rules for the branch.

GitBranchRule common fields

Present on every rule regardless of type .

Field	Type	Description
type	String	Rule type. See sections below for the full list.
rulesetSourceType	String	Source level: Repository, Organization, or Enterprise.
rulesetSource	String	Owner (for org), repo full name (for repository), or enterprise slug.
rulesetId	Long	GitHub ID of the ruleset this rule came from.

All remaining fields below are nullable; only the fields associated with a rule's type are populated, and the rest are null.

Rule type: pull_request

Field	Type	Description
requiredApprovingReviewCount	Int?	Minimum approvals required.
dismissStaleReviewsOnPush	Boolean?	Dismiss prior approvals when new commits land.
requireCodeOwnerReview	Boolean?	Require code-owner approval.
requireLastPushApproval	Boolean?	The most-recent push must be approved by someone other than the pusher.
requiredReviewThreadResolution	Boolean?	All review threads must be resolved before merging.
allowedMergeMethods	List<String>?	One or several of merge, squash, rebase.
requiredReviewers	List<RequiredReviewerConfiguration>?	(Beta) Reviewers required for changes matching specific file patterns.

Rule type: required_status_checks

Field	Type	Description
requiredStatusCheckContexts	List<String>?	Required check names.
strictRequiredStatusChecksPolicy	Boolean?	Branch must be up-to-date before merging.
doNotEnforceOnCreate	Boolean?	Allow repos/branches to be created even if a check would prohibit it. (Also populated for workflows.)

Rule type: required_deployments

Field	Type	Description
requiredDeploymentEnvironments	List<String>?	Required environment deployments.

Rule type: update

Field	Type	Description
updateAllowsFetchAndMerge	Boolean?	Branch can pull changes from its upstream repository.

Rule type: merge_queue

mergeQueueCheckResponseTimeoutMinutes	Int?	Max time a required check has to report a conclusion before being considered failed.
mergeQueueGroupingStrategy	String?	ALLGREEN or HEADGREEN.
mergeQueueMaxEntriesToBuild	Int?	Maximum queued PRs being checked/built simultaneously.
mergeQueueMaxEntriesToMerge	Int?	Maximum PRs merged together in one group.
mergeQueueMergeMethod	String?	MERGE, SQUASH, or REBASE. (Uppercase here; lowercase on allowedMergeMethods.)
mergeQueueMinEntriesToMerge	Int?	Minimum PRs in a merge group.
mergeQueueMinEntriesToMergeWaitMinutes	Int?	Minutes to wait before merging a smaller-than-minimum group.

Pattern rule types

Shared by commit_message_pattern, commit_author_email_pattern, committer_email_pattern, branch_name_pattern, tag_name_pattern. The same four fields are populated for any of these.

Field	Type	Description
patternName	String?	Display name for the pattern rule.
patternNegate	Boolean?	If true, the rule fails when the pattern matches.
patternOperator	String?	starts_with, ends_with, contains, or regex.
pattern	String?	The pattern to match against.

Rule type: workflows

Field	Type	Description
requiredWorkflows	List<WorkflowFileReference>?	Workflows that must pass for the rule to pass.
doNotEnforceOnCreate	Boolean?	Shared with required_status_checks . See above.

Rule type: code_scanning

Field	Type	Description
codeScanningTools	List<CodeScanningTool>?	Tools that must provide code-scanning results.

Rule type: copilot_code_review

Field	Type	Description
reviewDraftPullRequests	Boolean?	Copilot reviews draft PRs before they're marked ready.
reviewOnPush	Boolean?	Copilot reviews every new push.

Rule type: file_path_restriction

Field	Type	Description
restrictedFilePaths	List<String>?	File paths restricted from being pushed.

Rule type: max_file_path_length

Field	Type	Description
maxFilePathLength	Int?	Maximum number of characters allowed in file paths.

Rule type: file_extension_restriction

Field	Type	Description
restrictedFileExtensions	List<String>?	File extensions restricted from being pushed.

Rule type: max_file_size

Field	Type	Description
maxFileSize	Int?	Maximum file size in megabytes (excluding Git LFS).

Rule types without parameters

creation, deletion, required_linear_history, non_fast_forward, and required_signatures present in the rules list with only the common fields populated.

Nested types

RequiredReviewerConfiguration - each element of pull_request.requiredReviewers:

Field	Type	Description
filePatterns	List<String>	fnmatch patterns — PRs that change matching files require this reviewer.
minimumApprovals	Int	Required approvals from this reviewer. Zero makes approval optional but still adds the reviewer.
reviewer	RulesetReviewer	The reviewing team.

RulesetReviewer - nested in RequiredReviewerConfiguration.reviewer:

Field	Type	Description
id	Long	Reviewer ID.
type	String	Reviewer kind. Currently always Team.

WorkflowFileReference - each element of workflows.requiredWorkflows:

Field	Type	Description
path	String	Path to the workflow file in the repository.
ref	String?	Branch or tag ref of the workflow file to use.
repositoryId	Long	ID of the repository where the workflow is defined.
sha	String?	Commit SHA of the workflow file to use.

CodeScanningTool - each element of code_scanning.codeScanningTools:

alertsThreshold	String	Severity at which alerts block a push: none, errors, errors_and_warnings, all.
securityAlertsThreshold	String	Severity at which security alerts block: none, critical, high_or_higher, medium_or_higher, all.
tool	String	Name of the code-scanning tool.

Relationship with git.branchProtection()

git.branchProtection() returns the effective branch protection for a branch, merging policies from both classic branch protection and GitHub rulesets into a unified result. git.branchRulesets() remains available if you need granular access to individual ruleset rules (e.g. filtering by type, source, or ruleset ID).

Commits

Get the latest commits (to a maximum of 100) for a defined lookback period (defaulting to 7 days).

• Date

• Message

• SHA

• URL

• Username

These results can be filtered based on branch name, using the default branch if no other branch is provided.

Definition: git.commits()

Example

You can use the git.commits() expression in a security Scorecard to make sure entities have fewer than three commits to a "security-fixes" branch in the last week:

git.commits(branchName="security-fixes").length < 3

Entities passing this rule will include those that haven't needed three or more security fixes. This can indicate that there aren't vulnerabilities in a given entity's code, but could also suggest that fixes aren't being implemented. Using this rule in conjunction with one focused on vulnerabilities could provide the extra context needed to gain a better understanding of what's happening.

Default branch

Default branch for the repository, or main when null.

Definition: git.defaultBranch()

Example

If default branches should always be named "main," you can write a rule to make sure entities follow this practice:

git.defaultBranch().matches("main")

File contents

Load the contents of a file from the entity's associated repository.

The contents can be validated by using string comparison operations or parsed by the built-in jq function. The jq function will automatically coerce file contents of JSON or YAML formats.

Definition: git.fileContents()

Example

For a Scorecard focused on development maturity, you could use the git.fileContents() rule to enforce that a CI pipeline exists, and that there is a testing step defined in the pipeline.

git.fileContents("circleci/config.yml").matches(".*npm test.*")

A best practices Scorecard, meanwhile, could use this expression for a number of rules:

• To make sure node engine version in specified in the package.json file:

  jq(git.fileContents("package.json"), ".engines.node") != null

• To make sure TypeScript projects have a tsconfig.json file checked in:

  jq(git.fileContents("package.json"), ".devDependencies | with_entries(select(.key == \"typescript\")) | length") == 0 or git.fileExists("tsconfig.json")

• To make sure projects using yarn do not allow NPM:

  jq(git.fileContents("package.json"), ".engines.yarn") == null or jq(git.fileContents("package.json"), ".engine.npm") = "please-use-yarn"

• And to ensure the yarn version being used is not deprecated:

  jq(git.fileContents("package.json"), ".engines.yarn") == null or !(semver("1.2.0") ~= semverRange(jq(git.fileContents("package.json"), ".engines.yarn")))

File exists

Check if file exists from within the entity's associated repository.

Definition: git.fileExists()

Examples

For a Scorecard focused on best practices, you can make sure that repositories contain a README.md file:

git.fileExists("README.md")

This rule would make sense in the first level because it's so essential.

A higher-level rule in a best practices Scorecard might confirm that developers are checking in lockfiles to ensure consistency in package installs:

git.fileExists("yarn.lock") OR git.fileExists("package-lock.json")

And/or a rule that makes sure there are unit tests enabled:

git.fileExists("*Test.java")

Finally, you could write a rule to make sure projects have a standard linter:

git.fileExists(".prettierrc.json") OR git.fileExists(".eslintrc.js")

Number of Git vulnerabilities

Check the number of vulnerabilities for an entity's associated repository.

You can filter by severity (by default searches by all severities) or source (by default only searches GitHub security advisories).

When using the GitHub Advanced Security source, severities displayed in the UI may not match severities returned by the API.

Definition: git.numOfVulnerabilities()

Examples

A security-focused Scorecard will likely include a rule making sure there are no Git vulnerabilities:

git.numOfVulnerabilities() == 0

You can use Scorecard levels to stratify vulnerabilities by risk. An initial level might make sure there are no critical vulnerabilities:

git.numOfVulnerabilities(severity=["CRITICAL"]) == 0

While a higher level might make sure there are no vulnerability warnings:

git.numOfVulnerabilities(severity=["WARNING"]) == 0

List of Git vulnerabilities

Lists the vulnerabilities for an entity's associated repository. You can filter by severity (by default searches by all severities) or source (by default only searches GitHub security advisories). Note when using the GitHub Advanced Security source, severities displayed in the UI may not match severities returned by the API.

Definition: git.vulnerabilities()

Examples

You could write a Scorecard rule that verifies an entity has fewer than 5 Git vulnerabilities:

git.vulnerabilities().length < 5

You could write a rule that verifies the entity has no vulnerabilities with "High" or "Critical" severity sourced from GitHub Advanced Security:

git.vulnerabilities(severity=["CRITICAL", "HIGH"], source=["GITHUB_ADVANCED_SECURITY"]).length < 1

Has Cortex YAML

Check if a repository has a valid cortex.yaml file checked in at the root directory (when GitOps is enabled).

Definition: git.hasCortexYaml()

Example

If you're using a Scorecard to track a migration from Cortex UI to GitOps, you can use this rule to make sure entities are set up for GitOps management of entity descriptors:

git.hasCortexYaml() == true

Last commit details

Provides last commit details.

• Date

• Message

• SHA

• URL

• Username

Definition: git.lastCommit()

Examples

One of the first rules you might write for a Scorecard focused on development maturity or security is one validating that the last commit was within the last month:

git.lastCommit().freshness < duration("P1M")

As counterintuitive as it may seem, services that are committed too infrequently are actually at more risk. People who are familiar with the service may leave a team, institutional knowledge accumulates, and from a technical standpoint, the service may be running outdated versions of your platform tooling.

Depending on best practices at your organization, you may want to confirm entities are updated within a week:

git.lastCommit().freshness < duration("P7D")

Confirming whether a service was updated within the last week can help team members catch outdated code sooner. Plus, if there is a security issue, you can quickly determine which services have or have not been updated to patch the vulnerability.

Pull requests

Lists the pull requests opened during a defined lookback period. Bot-authored pull requests are included by default, so filter them out if you only want activity from people.

Each pull request has these properties:

Property	Description
approvalDate	When the pull request was approved. Returns null if not yet approved.
author	The GitHub user account or bot account that opened the pull request.
dateClosed	When the pull request was closed (merged or rejected). Returns null if still open.
dateOpened	When the pull request was opened. Always present.
firstReviewDate	When the pull request received its first review. Returns null if there are no reviews yet.
isBotGenerated	Whether a bot account opened the pull request.
isDraft	Whether the pull request is still a draft. Returns null if the data is unavailable.
labels	A list of label names applied to the pull request in GitHub (e.g. ["bug", "small", "stale"]). Returns null if no labels are present or the data is unavailable.
lastUpdated	When the pull request last changed.
numCommits	How many commits the pull request contains.
numLinesAdded	How many lines the pull request adds. Returns null if the data is unavailable.
numLinesDeleted	How many lines the pull request deletes. Returns null if the data is unavailable.
organization	The GitHub organization that owns the repository.
repository	The repository the pull request belongs to.
source	The Git provider the pull request came from, e.g. "GITHUB".
status	The current state of the pull request: "OPEN", "CLOSED", "MERGED", or "UNKNOWN".
title	The title of the pull request as set by the PR author in GitHub. Returns null if the title is unavailable.
url	A link to the pull request in GitHub.

Definition - git.pullRequests(lookback: Duration)

Example

Use the git.pullRequests() expression to find entities with very few pull requests opened in the last two weeks:

git.pullRequests(lookback=duration("P14D")).length < 3

This surfaces entities that nobody has touched lately, which is useful when you need to confirm that a fix, like a vulnerability patch, is actually being picked up.

Example

Filter out drafts to count only the pull requests that are ready for review:

git.pullRequests(lookback=duration("P14D")).filter((pr) => !pr.isDraft).length < 3

Example

Filter out bot accounts to count only the pull requests opened by people:

git.pullRequests(lookback=duration("P14D")).filter((pr) => !pr.isBotGenerated).length < 3

Reverse the filter to pr.isBotGenerated to look at automated activity on its own, for example to see how much of a repository's pull request volume comes from dependency bots.

Reviews

Lists the reviews left during a defined lookback period. Reviews left by bots are included by default, so filter them out if you only want reviews from people.

Each review has these properties:

Property	Description
isBotGenerated	Whether a bot account left the review.
organization	The GitHub organization that owns the repository.
repository	The repository containing the pull request that was reviewed.
reviewDate	When the review was left.
reviewer	The GitHub user account or bot account that left the review.

Definition - git.reviews(lookback: Duration)

Example

A development maturity Scorecard might use the git.reviews() expression to make sure that a rigorous review process is in place before changes are implemented:

git.reviews(lookback=duration("P7D")).length > 25

This rule makes sure that more than 25 reviews were left in the last week.

Example

Because bot reviews count toward that total, filter them out to hold the threshold to reviews left by people:

git.reviews(lookback=duration("P7D")).filter((review) => !review.isBotGenerated).length > 25

Reverse the filter to review.isBotGenerated to measure automated review activity on its own.

Workflow runs

Get workflow runs meeting given filter criteria, including conclusions, statuses, and a lookback period.

• Conclusion

• Name

• Run started at

• Run time

• Run updated at

• Status

Conclusions: FAILURE, SUCCESS, TIMED_OUT

Statuses: QUEUED, IN_PROGRESS, COMPLETED

The lookback period specifies a duration for which returned runs should be created within, defaulting to a period of 3 days.

• The runTime of the WorkflowRun object represents the difference between runStartedAt and runUpdatedAt times in seconds.

Definition: git.workflowRuns()

Example

To make sure an entity has had a successful workflow run within the last two weeks, you can write a rule like:

git.workflowRuns(conclusions=["SUCCESS"], statuses=["COMPLETED"], lookback=duration("P14D")).length > 0

This rule is checking for GitHub workflow runs with a SUCCESS conclusion and COMPLETED status during a 14-day lookback window.\

To find the percentage of successes in workflow runs, you could write a query similar to:

(git.workflowRuns(conclusions=["SUCCESS"], lookback=duration("P1D")).length) / (git.workflowRuns(lookback=duration("P1D")).length) * 100

Ownership CQL

All ownership details

A special built-in type that supports a null check or a count check, used to enforce ownership of entities.

Definition: ownership: Ownership | Null

Example

An initial level in a security Scorecard might include a rule to ensure an entity has at least one team as an owner:

ownership.teams().length > 0

All owner details

List of owners, including team members and individual users, for each entity

Definition: ownership.allOwners()

Example

The Scorecard might include a rule to ensure that entity owners all have an email set:

ownership.allOwners().all((member) => member.email != null)

Team details

List of teams for each entity

Definition: ownership.teams(): List<Team>

Example

The Scorecard might include a rule to ensure that an entity owners all have a description and are not archived:

ownership.teams().all(team => team.description != null and team.isArchived == false)

Copilot CQL

AI adoption

The ratio of licensed seats that were active users of AI coding tools in a given time period. Returns a value between 0 and 1, where 1 represents 100% adoption. Note: This metric is only available for Team entities.

Definition: aiTools.analysis(lookback: Duration).aiAdoptionRate

Example

You could create a Scorecard rule to make sure AI adoption rate is at least 50% over the previous 30 days:

aiTools.analysis(lookback = duration("P30D")).aiAdoptionRate >= 0.5

Active AI users

The number of users who used AI coding tools in a given time period. Note: This metric is only available for Team entities.

Definition: aiTools.analysis(lookback: Duration).activeAiUsers

Example

You could create a Scorecard rule to verify you had at least 5 active AI users in the last 7 days:

aiTools.analysis(lookback = duration("P7D")).activeAiUsers >= 5

External repositories

By default, each GitHub rule is evaluated on the repository defined in a given entity descriptor. If the base path parameter has been set, CQL rules will automatically scope to the base path subdirectory.

To evaluate the rule for a service for an external repository, pass the repo identifier in the git(repoIdentifier: Text) command (e.g. git("github:org/repositoryName")).

This can be combined with other CQL rules. For example, a rule based on a dynamic external repository with custom data would be git("github:" + custom("my-custom-repo")).fileExists("README.md").

View integration logs

This feature is available in Cortex cloud.

While viewing an integration's settings page, select the Logs tab to view error logs from the last 7 days. You can filter the logs list by configuration and by operation (for example, you could filter to view errors surfaced only via Scorecards).

Click into a row to get more information, including time stamp, status code, full error, and request path.

Background sync

Cortex syncs GitHub identities (teams and people) once a day, at 9 a.m. UTC.

Repository and pull request syncs start within 10 minutes of the previous sync's completion. In most cases, this means Cortex's GitHub data is no more than 10 minutes old, though a sync that takes longer than usual can push that slightly further out.

FAQs and troubleshooting

I'm getting this error: "{"message":"Not Found", "documentation_url":"https://docs.github.com/rest/repos#get-a-repository"}".

If you've set up multiple GitHub accounts/organizations, Cortex will not be able to identify the correct one unless the alias variable is defined.

What if I have multiple email addresses set in my GitHub account?

Cortex will only detect the primary email address associated with your GitHub account if it is public.

If Cortex is not correctly pulling in user emails, ensure the given user(s) have allowed their email address to be public. Make sure the "Keep my email address private" setting is unchecked in the user's personal GitHub settings.

My ownership isn't being automatically mapped through GitHub.

If the email address associated with your Cortex account is not the same as your GitHub email address, you need to add your Cortex email address to the Public email dropdown in GitHub settings.

Github OAuth, which you can configure in Cortex user settings, allows you to link your GitHub username with your Cortex account, even if you don't have a public email set up on GitHub.

Still need help?​

The following options are available to get assistance from the Cortex Customer Engineering team:

• Email: [e-mail], or open a support ticket in the in app Resource Center

• Slack: Users with a connected Slack channel will have a workflow added to their account. From here, you can either @CortexTechnicalSupport or add a :ticket: reaction to a question in Slack, and the team will respond directly.

Don’t have a Slack channel? Talk with your Customer Success Manager.