Intel / blog

How to continue a GitHub Copilot app session in VS Code

Source: github.blog
Author: Kayla Cinnamon
Date: Published Oct 19, 2026

Context

GitHub's tutorial explains a handoff between its Copilot app, where an does the bulk of a coding task, and VS Code, where a developer usually reviews, debugs and tests (GitHub Blog). Choosing Open in VS Code from a session opens the project on the same isolated worktree the agent used and shows the existing session in the Copilot sidebar. Three things come across together: the session , meaning the task, plan and conversation that produced the changes; the diff of everything the agent changed; and the worktree itself. GitHub frames this as continuing the session rather than rebuilding it, so there is no need to summarize the history into a new chat.

Because the conversation is preserved, follow-up prompts in VS Code can refer to earlier work, such as asking why a function was changed or asking for a test for a label just edited, without restating the task. Manual edits land in the same worktree, building directly on the agent's changes. And because VS Code opened the session's own branch, committing and pushing from the editor updates the pull request that was already tracking the session, without copying files into another checkout. The post also tells readers to treat the agent's code like any other proposed change: review the diff, run the project's tests and verify the behavior locally before pushing.

Key takeaways · AI-distilled

A GitHub Copilot app session can be opened directly in VS Code with its task, plan and conversation carried over alongside the agent's diff, so the work continues without re-describing it in a new chat.
VS Code opens the same isolated worktree the session used, so hand edits build directly on the agent's changes without copying files between directories or recreating the branch.
Pushing from that worktree updates the pull request already associated with the session, with no manual step to reconnect the work.
GitHub's guide says to review agent changes like a person's: check conventions, edge cases, unrelated edits and test coverage, then run the repository's tests and lint locally before pushing.

Terms in this piece · Glossary

AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.

Read the source github.blog