Skip to main content
AI Development

GitHub Copilot in Teams: switch apps before November 16

GitHub merged its Teams apps. The legacy GitHub app retires November 16, and channel requests to Copilot now open PRs under the app identity.

Sean McLellan profile photo

Sean McLellan

Lead Architect & Founder

5 min read
Diagram titled Who Copilot acts as in Teams. Two panels: in a direct message, an arrow points to your linked GitHub account; in a channel or group chat, an arrow points to the GitHub app identity, with a note reading one more approval before merge.
Constructed diagramConstructed diagram based on GitHub Docs, October 2026. Not GitHub product UI.

GitHub merged its Microsoft Teams apps on October 23, 2026: notifications and Copilot now live in one GitHub app for Teams. The legacy GitHub app will be renamed "GitHub (deprecated)" and retired on November 16, so teams that start Copilot work from Teams need the consolidated app installed before then.

The install takes minutes. The decision that takes longer is where you let people start Copilot sessions. In a direct message, Copilot acts with your own GitHub permissions. In a channel or group chat, it opens pull requests under its app identity, and by default GitHub then asks for one more approval before merge.

What changed with the GitHub app in Teams?

GitHub folded two apps into one. GitHub says the new app is the one "previously called GitHub notifications," and it now handles both channel notifications and @GitHub requests to Copilot. If you already run GitHub notifications, you update it to the latest version and your existing notifications keep working.

The legacy GitHub app is the one with a deadline. GitHub's changelog says that app "will be renamed 'GitHub (deprecated)' and retired on November 16," and that you should install the consolidated app "to continue using Copilot in Teams." After installing, you mention @GitHub and follow the prompts to connect your GitHub account and select a repository.

The workflow GitHub is pitching: a bug report lands in a Teams channel as a notification, someone asks @GitHub to investigate, Copilot fixes the bug and opens a pull request, and the team reviews and merges it. The Copilot cloud agent part of this is still in public preview, according to GitHub's Teams integration docs.

Diagram titled Moving to the consolidated GitHub app. Four steps connected by arrows: legacy app renamed GitHub (deprecated), install the consolidated GitHub app, @GitHub to connect your account, set a default repository per channel. A banner below reads: legacy app retired November 16.
Constructed diagramConstructed diagram based on GitHub's October 23, 2026 changelog. Not GitHub product UI.

Who does Copilot act as in a Teams channel?

It depends on where you ask. GitHub's docs draw a clear line:

Scroll sideways to see all 3 columns.

Comparison dimensionDirect message to the GitHub appChannel or group chat
Identity for pull requests and issuesYour linked GitHub personal accountThe GitHub app's identity
Context Copilot readsThe direct message only (GitHub's suggested way to limit context)The entire thread
Who can trigger changesYou, with write access to the repoParticipants with write access; anyone in the conversation can add input
Merge approvalsYour ruleset as configuredOne extra approval, on by default, when the ruleset requires at least one
Default repositoryNot availableSet per channel with @GitHub settings

Two rows deserve attention. First, context: Copilot "will capture the entire thread as context for your request," and that context "is stored in the artifacts the agent generates." A long incident thread with customer names or credentials pasted in can end up in a pull request description. GitHub's own advice is to use a direct message when you want to limit context.

Second, approvals. GitHub's rulesets documentation says the "additional approval for unattributed Copilot pull requests" setting is on by default for new and existing rulesets. A ruleset that requires one approval will require two for a channel-created Copilot pull request. The setting has no effect if your ruleset requires zero approvals.

Guests in the Teams workspace and outside collaborators on the repository can't start or steer a session.

What do you need before November 16?

Check four things before the cutover:

  • A paid Copilot plan for anyone starting sessions.
  • Copilot cloud agent and cloud sandboxes enabled by the organization or enterprise owner. GitHub notes that cloud sandbox policies share the same configuration as cloud agent policies.
  • Permission to install apps in the Teams workspace. GitHub's integration guide also lists Microsoft Public Developer Preview enabled for the Teams client as a prerequisite.
  • A budget for both meters. Sessions started in Teams consume AI credits, and cloud sandbox usage is billed separately. Our earlier post covers why a Copilot task from Teams can hit two budgets.

The app only needs to be installed once per team. After that, each member connects their own GitHub account the first time they mention @GitHub.

Default repositories decide where pull requests land

Each channel can have a default repository, and it matters more than it looks. If a channel has none, GitHub docs say Copilot "sets the repository you use in your first session in that channel as the channel's default repository." Whoever asks first picks the target for everyone after them.

Set it on purpose. Run @GitHub settings in each channel that will use Copilot and choose the repository. For one-off work against another repository, name it in the request: @GitHub Create a pull request to ... repo=OWNER/REPO_NAME branch=BRANCH_NAME.

Why it matters for teams that run work in Teams

BaristaLabs' interpretation: the retirement date is a forcing function, and that's useful. If Copilot in Teams has been running as a preview someone switched on, the migration is a cheap moment to decide how you want it used rather than inheriting whatever the old app was doing.

The channel model fits teams that triage bugs together. Everyone sees the investigation, and the extra approval keeps a person between agent-authored code and production. The cost is that the thread becomes part of the work record, and the first request in a channel can set its repository. Both are easy to manage if you set them before people start using the new app, and awkward to clean up afterward. The same thinking about keeping investigation separate from commitment applies here.

Our recommendation: this week, install the consolidated app in one team that already triages bugs in Teams. Set that channel's default repository, confirm your ruleset requires at least one approval so the extra approval applies, and post a one-line team rule about what not to paste into threads that might reach @GitHub. Then remove the legacy app from that team and repeat for the rest before November 16.

If you want a quick audit of which repositories allow app-created pull requests and how your approvals are set, our guide to reviewing ruleset bypasses with Rule Insights is a good companion, and our AI consulting team can help you plan the rollout.

Sources and scope

BaristaLabs did not test the consolidated app. Features, dates and defaults are as GitHub describes them, and the Copilot cloud agent integration is in public preview and may change. GitHub's changelog doesn't say whether the legacy app's notification subscriptions carry over to the consolidated app.

Copilot in Teams

Plan your move to the new GitHub app

Bring your Teams channels and your repository rulesets. BaristaLabs can help you pick default repositories, confirm the extra approval applies where it should, and write a short team rule for agent requests before November 16.

Start with the channel where your team triages bugs.

Turn this idea into a pilot

Which workflow should go first?

Use the readiness check to compare impact, effort, risk, owner, and next step before requesting a review.

  • 3-5 minutes
  • Deterministic score
  • No sensitive data
Check workflow readiness

Practical AI Workflow Notes

Want more practical AI operations ideas?

Get short notes on applying AI inside real small-business workflows — from document handling and customer follow-up to internal reporting, compliance, and automation guardrails.

A useful next step if you’re still exploring and not ready to request a 20-minute workflow assessment.

Occasional emails. Practical workflow guidance only. Unsubscribe anytime.