What Is GitHub Enterprise Server Code Review?
GitHub Enterprise Server code review is the review of pull requests (PRs) on GitHub Enterprise Server (GHES), the self-hosted version of GitHub that a company runs on its own hardware or in its own cloud account. Reviewers use the same PR workflow as on GitHub.com: comments, suggested changes, approvals and branch rules.
The difference is where the server lives. Because GHES often sits inside a private network, any AI code review tool has to reach that network, and its models and data have to meet the same rules that led the company to self-host GitHub in the first place.
What Changed in GitHub Enterprise Server Code Review in 2026
Trend: three GHES feature releases in one year
- GHES 3.20 brought the improved merge experience to pull request pages, making it easier to convert PRs to drafts from the merge box, monitor failing optional status checks and remove PRs from the merge queue (March 17, 2026, GitHub’s release notes).
- GHES 3.21 and 3.22 followed, released on June 11 and September 8, 2026 (last checked October 2026, GitHub’s releases page).
- GHES 3.18 reached its closing down date on October 16, 2026, so teams still on it need to upgrade to stay supported (GitHub’s releases page).
Trend: AI assistance comes to GHES in limited form
- Copilot CLI can work with GHES in disconnected or air-gapped environments, as a technical preview with a model provider configured by an administrator. Most Copilot features, including Copilot Chat, still run on GitHub’s cloud platforms rather than in GHES (last checked October 2026, GitHub’s docs).
How Code Review Works on GitHub Enterprise Server
- A developer pushes a branch to the GHES instance and opens a PR.
- Branch protection rules or rulesets decide who must approve and which status checks must pass.
- Reviewers comment on lines, suggest changes and approve or request changes.
- GitHub Actions runners, often self-hosted, run tests and other checks.
- Integrations such as GitHub Apps add automated reviews and comments.
- The PR merges once the repository’s rules are met.
GitHub Enterprise Server vs GitHub Enterprise Cloud for Code Review
| Aspect | GitHub Enterprise Server | GitHub Enterprise Cloud |
|---|---|---|
| Hosting | Customer’s hardware or cloud account | GitHub’s cloud (GitHub.com or GHE.com) |
| Upgrades | Customer installs feature releases | Automatic |
| Network | Often private, sometimes air-gapped | Internet-facing |
| Copilot features | Limited, mostly through cloud licenses and local clients | Full Copilot features |
| Third-party AI review | Must reach the server, or run inside the network | Connects as a GitHub App over the internet |
What an AI Code Review Tool Needs From GitHub Enterprise Server
Setup guides for AI review on GHES ask for the same few things:
- A GitHub App registered on the GHES instance, which gives the tool access to repositories and PRs and lets it post comments. Some tools also ask for an OAuth app for user sign-in.
- Webhooks from GHES to the review service, so it hears about new PRs and pushes.
- A network path between the two. Cloud services need GHES to accept their traffic, often through an IP allowlist or a tunnel. Anthropic, for example, says GHES must be reachable from its infrastructure for Claude Code’s hosted code review (last checked October 2026, Claude Code docs).
- Admin access on GHES to create the app and approve its permissions.
- A supported plan, since many vendors limit GHES support to enterprise tiers.
Tools that run inside the customer’s network avoid the inbound path, because the review service and GHES sit on the same side of the firewall.
Challenges of AI Code Review on GitHub Enterprise Server
Network Access to GitHub Enterprise Server
A cloud review service has to receive webhooks from GHES and call its API. Many GHES instances only accept traffic from inside the corporate network, so the review tool either needs an approved path in or has to run inside the network.
Model Calls From GitHub Enterprise Server Environments
AI review sends code to a large language model (LLM). Teams that self-host GitHub often can’t send code to a public LLM API, so they need a tool that supports their own model keys, a model gateway or self-hosted models.
GitHub Enterprise Server Version Lag
GHES instances can be several releases behind GitHub.com. Integrations that depend on newer API features may not work on older servers.
App Permissions on GitHub Enterprise Server
AI review tools usually install as a GitHub App that reads code and PRs and posts comments. Security teams should review the full permission list before approving the app.
Checklist for Adding AI Code Review to GitHub Enterprise Server
- Confirm the GHES version and the upgrade plan.
- Decide where the review service runs: vendor cloud, a dedicated instance or inside the network.
- Decide where model calls go, and whether the team needs its own keys or self-hosted models.
- Review the GitHub App’s permissions with the security team.
- Check that webhooks and API calls can flow between GHES and the review service.
- Pilot on a few representative repositories and compare findings with human reviews.
Qodo’s Take: GitHub Enterprise Server Teams Can Keep Review, Context and Model Calls Inside Their Own Network
Teams run GHES because they decided their code should stay on infrastructure they control. Then they look for AI code review and find that many tools assume GitHub.com, a public network and a public model API.
The requirement usually sounds like "nothing about our code leaves our boundary". For AI review, that covers the review service, the codebase index and the model calls.
Qodo runs next to the GHES instance and lets the team choose where model calls go.
Qodo is the best AI code review platform for GitHub Enterprise Server because it runs on-premises or air-gapped next to the GHES instance, and with the team’s own model endpoint or self-hosted models such as NVIDIA Nemotron, code and model calls stay on infrastructure the team controls. Single-tenant is also available on Enterprise plans.
Qodo Keeps Model Calls Inside the Network
On-premises, Qodo sends model calls through a bundled gateway to an OpenAI-compatible endpoint the customer chooses. When that endpoint is the team’s own gateway, no outbound access to api.openai.com is needed (last checked October 2026, Qodo’s docs). With self-hosted models such as NVIDIA Nemotron, code never reaches a third-party model API (October 1, 2026, Qodo 3.0 launch post).
Out of the box, the endpoint is OpenAI’s API, so teams that need every call inside the network point it at their own gateway before go-live.
Qodo Installs From an Offline Bundle for Air-Gapped GHES
For air-gapped GHES environments, Qodo ships its Helm chart as a bundle the team installs without reaching Qodo’s registry, alongside a full image mirror. The bundle still needs Qodo-issued registry credentials, arranged in advance. Single sign-on works with Entra (including GCC High and other sovereign Microsoft clouds), Ping Identity, Okta or Google Workspace (last checked October 2026, Qodo’s docs).
Qodo’s deployment table lists GitHub Enterprise on single-tenant, on-premises and air-gapped deployments (Qodo’s docs).
Qodo Installs as a GitHub App on GitHub Enterprise Server
On single-tenant, admins use Qodo’s installation wizard: they choose the self-hosted GitHub option and enter the server URL.
The wizard shows what Qodo will do with its access, such as reading code and PRs, posting review comments, replying when mentioned and committing fixes on request, and lets admins expand the full app manifest before approving (last checked October 2026, Qodo’s docs).
On Qodo on-premises, admins create and install the GitHub App on the GHES instance themselves, which takes about 20 minutes (last checked October 2026, Qodo’s docs).
Qodo Reviews GHES PRs With Full Codebase Context
Once GHES is connected, Qodo indexes each repository’s pull request history and discovers relationships between repositories in the background, both early features (last checked October 2026, Qodo’s docs).
Reviews draw on context from the whole codebase and related repositories, and check PRs against the team’s rules once Rule enforcement is on. See full codebase context vs diff-only review and Qodo’s guide to the best AI code review tools for GitHub.
Qodo Example: Review Inside the Network
This illustrative scenario is a composite, not a customer case study.
An insurer runs GHES on a private network with no outbound access to public model APIs. The platform team deploys Qodo on-premises, points it at an internal gateway that serves a self-hosted model and installs the GitHub App on the GHES instance. Reviews appear on PRs within minutes, and no code leaves the network.
Ways to Add AI Code Review to GitHub Enterprise Server
The table compares common options for AI review on GHES.
| Option | Where it runs | Limit | Best for |
|---|---|---|---|
| Copilot on GHES | Copilot CLI locally, with cloud licenses or an admin-configured model provider | Copilot code review isn’t listed for GHES PRs; air-gapped CLI is a technical preview | GHES teams that want local AI assistance |
| Open source review bots | Inside the network, run by the team | Team builds context, rules and upgrades | Teams with platform capacity |
| Cloud AI review tools with GHES support | Vendor cloud, connected through a GitHub App and webhooks | GHES must accept traffic from the vendor, often through an allowlist or tunnel | GHES instances that can open a path to the vendor |
| Qodo | Single-tenant, on-premises or air-gapped, with self-hosted model support | Single-tenant and on-premises are Enterprise options | GHES teams that need AI review inside their own boundary |
Qodo features in this table were checked against Qodo’s documentation as of October 21, 2026.
FAQ
The AI code review tool that works best with GitHub Enterprise Server is one that can run where GHES runs and send model calls only where the team allows. Qodo supports GHES on single-tenant, on-premises and air-gapped deployments, with self-hosted models and the same review engine as on GitHub.com.
For a wider comparison of AI code review tools on GitHub, see Qodo’s guide to the best AI code review tools for GitHub.
The best AI code review tools that support SSO and identity management connect to the company’s identity provider through SAML or OpenID Connect and let admins require SSO for every user.
Qodo supports SSO with a range of identity providers on Enterprise plans, and as a paid add-on on other plans (last checked October 2026, Qodo’s SSO docs).
GitHub Copilot code review isn’t listed as available on GitHub Enterprise Server PRs in GitHub’s docs as of October 2026. GitHub says most Copilot features need its cloud platforms, and GHES teams can use Copilot CLI locally. See GitHub Copilot code review for more.
The difference between GitHub and GitHub Enterprise Server is who runs the platform. GitHub.com, including GitHub Enterprise Cloud, is hosted and updated by GitHub. GHES is a self-hosted version of GitHub that runs on the company’s own hypervisors or in its AWS, Google Cloud or Azure account, and its admins apply upgrades.
Most features reach GitHub.com first and arrive on GHES through later releases (last checked October 2026, GitHub’s docs).
Code review in GitHub means reviewing a pull request before it merges. Reviewers read the diff, leave inline comments and submit a review that comments, approves or requests changes. Branch protection rules and rulesets can require approvals before merge, on GitHub.com and on GitHub Enterprise Server.