Atlassian is expanding Jira and Rovo with Code Context and agent loops. The company aims to move AI agents from isolated sessions to continuously running, managed workflows.
Atlassian states that almost no one can run agents at scale without things breaking, and the new features are designed to address that. The key to this is context. According to Atlassian, agents fail because they don’t understand an organization’s architecture, decisions, and standards. That knowledge lives in team members’ minds, in documentation, and in chat channels. Code Context, built on the Teamwork Graph, gives Rovo and coding agents insight into codebases spread across multiple repositories. This insight is intended to cover the entire lifecycle: from assessing the feasibility of backlog ideas to bug triage and root-cause analysis.
In addition, Agent Context Controls are being introduced. Platform teams use these to determine which agents can operate in a space and exactly what they can see.
From one-off prompts to continuous loops
The second component focuses on execution. Agent loops in Jira continuously scan for well-defined, unassigned work items, delegate them to the Jira Coding Agent, and then open a pull request that’s ready for review. Standards allows platform teams to define coding guidelines once and link them to repositories. AI Review adds an agent to every pull request, which checks against those guidelines before a human reviews it.
“You stay in control of the merge button but stop being the bottleneck for everything leading up to it,” Atlassian writes about the result.
The third problem Atlassian identifies: no one can prove that it works. DX for Agentic Development measures AI’s impact on throughput, quality, adoption, and costs, and links expenses to delivered work. This includes AI Code Insights, tool and MCP tracking, model-to-task fit, and research into Agent Experience. The Jira Agent Usage Dashboard shows team leaders which agents are being used in which workflows.
Tip: Atlassian is giving Jira and Rovo Agents an “agentic” upgrade