The GitHub Copilot coding agent has expanded its capabilities to now support resolving merge conflicts directly on pull requests. Announced March 26, 2026, this capability extends GitHub’s AI-powered development workflow from code generation and review assistance to maintenance tasks that traditionally require manual developer intervention. The feature represents a meaningful step toward fully agent-assisted development workflows where routine maintenance is increasingly automated.
How AI-Powered Conflict Resolution Works
Requesting AI-powered conflict resolution is straightforward. Developers simply mention @copilot in any pull request comment with specific instructions describing what they need. Common usage patterns include standard merge operations like @copilot Merge in main and resolve the conflicts or more detailed requests like @copilot Resolve merge conflicts and keep my changes for the authentication logic while accepting upstream changes for the database layer.
The Copilot agent operates within a secure cloud-based development environment provisioned specifically for the task. From this environment, it can fetch the latest state of the base branch, perform the merge operation, intelligently resolve conflicts using its trained understanding of both codebases involved, execute any configured builds and tests to validate the result, and push the resolved state back to the pull request branch. The entire process is transparent.
What This Changes for Development Workflows
Merge conflicts are an inevitable occurrence in collaborative software development, happening when parallel branches modify overlapping regions of code. On active repositories with multiple contributors working simultaneously, keeping long-running feature branches current can require repetitive manual effort that interrupts development flow. Developers must context-switch away from their current task, review changes made in the base branch, and carefully integrate them without introducing regressions.
By delegating this tedious task to Copilot, development teams can substantially reduce branch staleness, which is the problematic situation where feature branches diverge so far from main that eventual integration becomes painful and risky. The AI’s demonstrated ability to understand code semantics means it can make context-aware choices during conflict resolution rather than blindly accepting one version or the other.
Broader Agent Capabilities
The merge conflict resolution capability builds upon existing Copilot coding agent functionality that has been rolling out over recent months. Developers can already request help with fixing failing GitHub Actions workflows by commenting @copilot Fix the failing tests after observing a CI failure. They can address code review feedback with simple responses like @copilot Address this review comment, or make arbitrary changes with specific natural language requests.
These capabilities share a common architecture where the agent operates in an isolated environment with access to the repository and follows instructions expressed in natural language. The agent understands the repository context, the pull request context, and the specific instructions provided.
Availability and Configuration
The Copilot coding agent feature is available to customers on all paid Copilot plans, including Copilot Pro for individual developers, Copilot Business for team environments, and Copilot Enterprise for large organizations. For business and enterprise customers, organization administrators must explicitly enable the coding agent feature in organization settings before individual team members can utilize it.
Once enabled, the @copilot mention works in any pull request or issue comment. The agent acknowledges the request with a status comment that updates as work progresses, providing transparency into the agent’s activities.
Important Considerations
While this capability is genuinely promising for reducing maintenance burden, AI-driven conflict resolution introduces workflows that teams should evaluate thoughtfully. Agent-generated resolution commits should receive the same careful scrutiny as any other code change through established pull request review processes. Automated test coverage becomes more critical when human developers are not manually verifying each conflict resolution decision.
Sources
Ask @copilot to resolve merge conflicts on pull requests – GitHub Changelog