GitHub brings the Copilot coding agent to Jira in public preview

GitHub announced on March 5, 2026 that Jira issues can now be assigned directly to the GitHub Copilot coding agent in public preview. Teams can assign an issue to Copilot or @mention the agent in Jira comments, and GitHub will generate a draft pull request in the connected repository.

What the integration does

According to GitHub, the Copilot coding agent reads the Jira issue description and comments to gather context, works independently to implement changes, opens a draft pull request, and posts updates back into Jira through the agent panel. When the specification is incomplete, the agent can also ask clarifying questions directly in Jira.

The feature is aimed squarely at reducing the handoff friction between issue tracking and code execution. GitHub’s launch post highlights repetitive tasks such as bug fixes and documentation updates, where the overhead of moving between Jira and GitHub can be larger than the change itself.

Requirements and controls

Using the integration requires installing the GitHub Copilot for Jira app from the Atlassian Marketplace, which also installs a GitHub app as part of setup. Teams also need Jira Cloud with Rovo enabled, Copilot coding agent access, and a connected GitHub repository configured for the agent.

GitHub says the pull requests created by Copilot still follow the organization’s existing review and approval rules. That design matters because it positions the agent as an execution layer inside existing governance, rather than as a separate side channel that bypasses repository policy.

Why it matters

This release expands coding agents beyond the IDE or terminal and into the system where many teams define work in the first place. If the workflow holds up in practice, it could make issue description, implementation, and status reporting feel more continuous across PM and engineering tools. It also puts more weight on the quality of the Jira ticket itself, because weak specifications will now flow more directly into AI-generated code.

The feature is still in public preview, so the main question is not whether Copilot can produce a draft pull request, but whether teams can integrate that behavior into real delivery processes without creating review noise, permission sprawl, or brittle automation.

Source: GitHub