In this era of AI-assisted software development, developers need to know what to build and how to govern it, while coding agents need context to understand how to execute correctly.
To help organizations navigate and succeed with AI-native development and delivery, Atlassian today is releasing a new set of capabilities in Jira that the company said effectively create a context-rich orchestration layer for autonomous coding agents
Atlassian added these capabilities to address the gap between how much code AI is generating and the lack of productivity gains by developers. Among the issues the industry faces with implementing AI successfully are a lack of context that causes agents to drift from requirements, prompts that have no memory so prior work has to be redone, and a lack of governance over autonomous agents.
“When the customer doesn’t feel like they have to learn a completely new set of things, but rather with their knowledge of the existing Jira, and that we put those new features in the place where they can easily discover and use them, the concept should be intuitive,” Ming Wu, Head of Engineering, DevAI, at Atlassian, explained to SD Times.
Among the new capabilities in Jira are Jira for Slack, which enables teams to create context-rich specifications from conversations, feedback and ideas using @Jira. According to Atlassian’s announcement, “the agent updates work items, syncs conversations as comments, and assigns work to coding agents while your team collaborates in Slack.”
WIth this release, the company is introducing Jira Planner for spec-driven development. Jira Planner gathers up code pulls, the team’s Jira and Confluence history as well as team context to create requirements. Then, it can generate a spec in Confluence that developers or agents can build on. Further, work items can be assigned to models and agents such as Claude Code, Cursor or GitHub Copilot directly from within Jira, providing the context to get better responses from coding agents.
Additionally, video meetings can be turned by Atlassian’s Loom video messaging software into instructions and action plans agents can use to work on tasks. It’s these contextual assets that allow the agent to perform well, Wu said. “Context engineering is not just giving you the raw data. It’s the efficient way to retrieve the right context for your agent,” she said. “More context doesn’t necessarily mean better. With Jira Planner, you can go start from Jira and do the planning work with your team. And during the planning phase, one of the key things is putting all the contacts together from everywhere. We’re tryingto naked that process super convenient and also effective, making sure the right context surfaces during the planning stage.”
To get total visibility into agent behavior, Atlassian’s Teamwork Graph collects session records accessbile from anywhere in Jira, the company announced, along with new hooks in the Teamwork Graph CLI that can link local agent sessions directly to work in Jira, updating context continuously to avoid agent drift.
According to Atlassian, Jira for Slack, Jira Coding Agent, Jira agent automations, agentic templates, and agent sessions in Jira are available today for paid Jira Cloud customers at no additional cost. Jira Planner is available in early access, and Codex in Jira is coming soon. DX AI cost management is available for Atlassian DX customers.
