GitHub Copilot App Adds Stacked Sessions and Pull Requests for Sequential Coding Tasks

GitHub has published a blog post detailing a workflow feature called stacked sessions and pull requests in the GitHub Copilot app, aimed at developers handling multi-step coding tasks such as codebase modernization.

The post, written by a GitHub team member, walks through a real-world use case: modernizing an aging codebase using the feature. According to GitHub, stacking allows a developer to build a sequence of Copilot sessions and their resulting pull requests on top of one another, rather than handling each change as an isolated, one-off task.

What's confirmed

The source material is a narrative walkthrough rather than a formal feature announcement with a changelog, so several specifics — such as exact UI mechanics, availability (general availability vs. preview), supported plans, and pricing — are not detailed in the available content. GitHub has not disclosed pricing changes tied to this workflow, and no benchmark or performance data accompanies the post.

What is confirmed: the GitHub Copilot app now supports a stacking model for sessions and pull requests, and GitHub is promoting this as a practical approach for iterative, multi-stage code changes — the kind of work involved in migrating legacy code, refactoring across multiple files, or applying a chain of dependent changes where each step builds on the last.

Why stacking matters for legacy code work

Codebase modernization typically involves sequential dependencies: upgrading a dependency might require updating call sites, which might require adjusting tests, which might require further refactoring. Handling each of these as a disconnected Copilot session and separate PR creates friction — context has to be re-established at every step, and reviewers must piece together how changes relate to each other.

A stacked approach lets each session and PR build directly on the previous one, preserving context across the chain and giving reviewers a linear history of how a large change was decomposed. This mirrors stacked diff workflows already common in engineering teams that use tools like Graphite or Sapling for managing dependent pull requests, suggesting GitHub is bringing a familiar pattern from manual engineering practice into an AI-assisted context.

What this means

This is a workflow and UX update to the Copilot app rather than a new model release — there's no new model, context window, or benchmark attached to this change. Its value will depend on execution details GitHub hasn't fully spelled out in this post: how conflicts between stacked sessions are resolved, how review tooling surfaces the dependency chain, and whether the feature is available to all Copilot tiers or gated to specific plans.

For teams doing large-scale refactors or legacy modernization, a native stacking workflow inside Copilot could reduce the overhead of manually re-establishing context for each AI session. But until GitHub publishes fuller documentation — including availability and any plan restrictions — engineering teams should treat this as an emerging capability worth testing rather than a fully specified feature.