Ontology-guided extraction · Reviewed reference

GitHub product features

A bounded crawl of 14 official documents supports a reference graph of 13 GitHub.com capabilities. Open a feature to inspect the evidence behind its classification.

GitHub.comhas feature →13 feature nodeseach has →one assessed class

Feature classifications

AI features

5

The documented operation uses AI.

GitHub Copilot code review

The review is generated by Copilot and the documented mechanism uses agentic capabilities and model/token usage. Automatic review is a configured mode of the same reviewer.

Each time {% data variables.product.prodname_copilot_short %} reviews a pull request or reviews code in your IDE, the interaction consumes {% data variables.product.prodname_ai_credits_short %}. The amount depends on the model used and the number of tokens processed.
Pinned source · line 105
Copilot pull request summaries

The source explicitly calls the generated pull request summary AI-powered. Summarizing changes is distinct from reviewing or approving them.

intro: 'Generate an AI-powered summary of your pull request changes to help reviewers quickly understand what you changed and why.'
Pinned source · line 4
Agent Merge

Copilot performs remedial work after a configured blocker occurs. This capability and the native merge action are separately enabled.

When one of the blockers you selected appears, {% data variables.product.prodname_copilot_short %} starts working in a cloud development environment to fix it.
Pinned source · line 38
Copilot cloud agent

The agent is defined as an AI system. The scoped GitHub cloud workflow delegates repository research, changes and a pull request to it.

## What is an agent? An agent is an AI system that can complete a goal on your behalf. Unlike an assistant that only returns a response, an agent can plan a task, use tools to take action, inspect the results, and adjust its approach. For example, when you ask an agent to fix a bug, it can search a repository, identify the relevant files, edit the code, run tests, and revise its changes in response to test failures. The tools and environments available to an agent depend on where you use it.
Pinned source · lines 12–16
Copilot automations

The trigger can follow rules, while the named Copilot automation executes the AI agent and can use a chosen model. Ordinary automation is a separate concept.

intro: '{% data variables.copilot.copilot_automations_cap %} let you run {% data variables.copilot.copilot_cloud_agent %} automatically on a schedule, in response to events in a repository, or programmatically via the API.'
Pinned source · line 5

Non-AI features

7

The documented operation is human-controlled, static, or rule-driven.

Human pull request reviews

The documented actor in this scoped review mode is a person. It remains distinct from an approving review submitted by Copilot under an optional policy.

Pull request reviews let people comment on changes, suggest improvements, and approve or request changes before code is merged. Anyone with read access can review and comment on proposed changes, helping teams maintain code quality and share knowledge.
Pinned source · line 21
Pull request auto-merge

The stated behavior is a rule-triggered merge. Related Copilot blocker remediation may run first without changing this merge rule.

{% data variables.copilot.agent_merge %} fixes blockers, while auto-merge merges a pull request after all required reviews and status checks pass. Enabling {% data variables.copilot.agent_merge %} does not enable auto-merge.
Pinned source · line 66
Branch protection rules

The documented rules enforce selected push, merge, force-push or delete conditions. They can gate on separate features without inheriting how those features produce results.

By default, each branch protection rule disables force pushes to the matching branches and prevents the matching branches from being deleted. You can optionally disable these restrictions and enable additional branch protection settings.
Pinned source · line 35
Required status checks

The gate checks allowed status values or a configured status provider. This classification does not characterize the mechanisms used by external check providers.

Use these checks when a required status check blocks merging or pushing to a protected branch. See [AUTOTITLE](/pull-requests/reference/status-checks). * A required status check must have completed successfully in the chosen repository during the past seven days.
Pinned source · lines 19–21
Dependabot security updates

The source specifies the vulnerable-dependency trigger, compatibility against the dependency graph and the minimum patched version. The neighboring AI-agent assignment workflow is out of this capability's scope.

{% data variables.product.prodname_dependabot %} checks whether it's possible to upgrade the vulnerable dependency to a fixed version without disrupting the dependency graph for the repository. Then {% data variables.product.prodname_dependabot %} raises a pull request to update the dependency to the minimum version that includes the patch and links the pull request to the {% data variables.product.prodname_dependabot %} alert, or reports an error on the alert. For more information, see [AUTOTITLE](/code-security/reference/supply-chain-security/troubleshoot-dependabot/dependabot-errors).
Pinned source · line 47
Dependabot version updates

The documented mechanism reads configured manifests and compares versions. Keeping versions current is distinct from patching known vulnerable dependencies.

For each action in the file, {% data variables.product.prodname_dependabot %} checks the action's reference (typically a version number or commit identifier associated with the action) against the latest version. If a more recent version of the action is available, {% data variables.product.prodname_dependabot %} will send you a pull request that updates the reference in the workflow file to the latest version.
Pinned source · line 62
GitHub Pages

The source describes static hosting of specified files with an optional build step. The classification covers this publishing mode, not arbitrary contents or user-supplied build tools.

{% data variables.product.prodname_pages %} is a static site hosting service that takes HTML, CSS, and JavaScript files straight from a repository on {% data variables.product.github %}, optionally runs the files through a build process, and publishes a website. You can see examples of {% data variables.product.prodname_pages %} sites in the [{% data variables.product.prodname_pages %} examples collection](https://github.com/collections/github-pages-examples).
Pinned source · line 26

Needs review

1

The feature is documented. Its assessed mechanism remains open.

GitHub Copilot usage metrics dashboard

The source establishes an instrumentation and dashboard feature that measures use of AI-assisted workflows. It does not establish AI execution or a decisive manual or rule-based classification mechanism for this scoped dashboard.

* The usage metrics dashboard shows adoption and engagement trends.
Pinned source · line 32

Directed relationships

Reproduce the inventory

OpenSearch 2.19 variant

Use a separate documentation crawl to assess seven scoped operations. Local and provider-hosted embeddings execute a model. Keyword ranking, supplied-vector retrieval, score normalization, and request dispatch have fixed documented behavior in this scope. The private endpoint's implementation remains open.

Power Automate cloud flows

Assess authoring assistance, configured flow execution, external service mechanisms, and administrative controls against a fixed ontology.

Source revision f309096a83abde7c311523f8ca97fc587388b0e1. Quotes retain the source Markdown, including Liquid variable references. Documentation by GitHub Docs contributors, CC BY 4.0.