Debug Across Frontend and Backend Repositories with Copilot in Slack
Customers see an empty cart after signing in, and nobody knows whether the frontend or the API is at fault. How does Copilot in Slack trace it across both repositories?
- Ask
@GitHubto investigate frontend and backend code in one Slack conversation. - Omitted repository and branch names fall back to the channel's defaults.
- The integration is in public preview; app access and repository permissions still apply.
Follow the bug across the UI and API
A split frontend/backend codebase no longer requires separate Slack conversations for a cross-repository investigation. GitHub's October 15 update explicitly includes investigating an issue using code from both repositories.
For a hypothetical empty-cart bug after sign-in, a request could look like this:
“@GitHub Investigate why the cart appears empty after sign-in, using my-org/web and my-org/api. Compare the frontend request with the API response fields. Return likely causes, relevant files and missing reproduction details. Investigate only for now; do not change code or open a PR.”
Replace the names and symptom with the actual task. Review the findings before choosing which repository should receive a fix. Cross-repository context does not promise an atomic change or deployment across both projects.
An omitted name invokes the channel default
The default-repository setting explains why a request can land in an unexpected project. Without explicit repository or branch names, Copilot uses the channel's default repository and its default branch. The first session establishes a repository default if none exists.
Send @GitHub settings to change it. That change affects everyone in the channel, so naming the target in a one-off request is often clearer. Direct messages do not support the channel default-repository setting.
Move the discussion forward with the task
The full thread supplies context. Keep reproduction steps, expected results and observed behavior together; use a fresh thread or DM when the original discussion includes unrelated work.
The Slack product page describes dedicated Code Channels that keep a session's plan, branch and pull request together. Teammates can contribute context and follow progress there. Once a Code Channel exists, continue steering in that channel.
For an investigation, assess the proposed cause and supporting files. If the task moves to implementation, assess the actual diff and checks. Participation in the conversation does not confer authority to approve or merge changes.
Connect the app and verify repository access
As of October 26, the current integration documentation lists paid Copilot plans and public-preview status. The user starting cloud-agent work needs write access to the target repository. For enterprise repositories, the app must also be configured to access the relevant repositories. Workspace guests and repository outside collaborators cannot start or steer Slack sessions.
Use the GitHub app setup to install the integration and connect your account. Cloud sandboxes and cloud-agent policies must allow the work; an organization owner may need to enable them.
Then start with both repository names and a concrete symptom. The useful handoff is an investigation grounded in the relevant code, followed by an explicit decision about where a change belongs.