Draft poster · the finished cut will appear here
What happened
Atlassian describes a plan to turn well-defined, unassigned Jira work into an agent loop. The company says Jira can scan for eligible work, delegate it to Jira Coding Agent for implementation and testing, and open a pull request ready for review. Developers and product managers still control the merge decision.
How it works
The loop is surrounded by context and policy. Atlassian says Code Context, built on the Teamwork Graph, gives Rovo and coding agents secure information across multiple repositories. Agent Context Controls determine which agents can operate in a space and what they can see. Standards maps organizational coding rules to repositories, while AI Review checks pull requests against those rules before a human review.
Atlassian also describes DX for Agentic Development as an instrumentation layer for throughput, quality, adoption, and cost. That makes the proposed system more than a code-generation button: it connects task eligibility, repository knowledge, execution, verification, approval, and measurement.
Why it matters
Once agents can write patches, the bottleneck moves to context, policy, verification, and knowing whether the automation helps. An issue tracker and a Git workflow already contain much of the state an agent needs; governance decides where the loop is allowed to run.
What we don’t know yet
Atlassian says Code Context is rolling out in open beta, while agent loops, Standards, and AI Review are in private early access. Other controls are planned for later. Availability and productivity claims should not be treated as general release or independent proof of better engineering.
Source: Atlassian Blog, Taroon Mandhana, September 10, 2026 - original post