On this period of AI-assisted software program improvement, builders have to know what to construct and tips on how to govern it, whereas coding brokers want context to grasp tips on how to execute accurately.
To assist organizations navigate and succeed with AI-native improvement and supply, Atlassian at the moment is releasing a brand new set of capabilities in Jira that the corporate mentioned successfully create a context-rich orchestration layer for autonomous coding brokers
Atlassian added these capabilities to handle the hole between how a lot code AI is producing and the dearth of productiveness beneficial properties by builders. Among the many points the trade faces with implementing AI efficiently are a scarcity of context that causes brokers to float from necessities, prompts that don’t have any reminiscence so prior work needs to be redone, and a scarcity of governance over autonomous brokers.
“When the client doesn’t really feel like they should be taught a totally new set of issues, however slightly with their data of the present Jira, and that we put these new options within the place the place they’ll simply uncover and use them, the idea ought to be intuitive,” Ming Wu, Head of Engineering, DevAI, at Atlassian, defined to SD Instances.
Among the many new capabilities in Jira are Jira for Slack, which allows groups to create context-rich specs from conversations, suggestions and concepts utilizing @Jira. In keeping with Atlassian’s announcement, “the agent updates work objects, syncs conversations as feedback, and assigns work to coding brokers whereas your workforce collaborates in Slack.”
WIth this launch, the corporate is introducing Jira Planner for spec-driven improvement. Jira Planner gathers up code pulls, the workforce’s Jira and Confluence historical past in addition to workforce context to create necessities. Then, it may possibly generate a spec in Confluence that builders or brokers can construct on. Additional, work objects might be assigned to fashions and brokers comparable to Claude Code, Cursor or GitHub Copilot straight from inside Jira, offering the context to get higher responses from coding brokers.
Moreover, video conferences might be turned by Atlassian’s Loom video messaging software program into directions and motion plans brokers can use to work on duties. It’s these contextual belongings that permit the agent to carry out effectively, Wu mentioned. “Context engineering is not only providing you with the uncooked knowledge. It’s the environment friendly strategy to retrieve the suitable context in your agent,” she mentioned. “Extra context doesn’t essentially imply higher. With Jira Planner, you may go begin from Jira and do the planning work along with your workforce. And throughout the planning part, one of many key issues is placing all of the contacts collectively from in every single place. We’re tryingto bare that course of tremendous handy and in addition efficient, ensuring the suitable context surfaces throughout the starting stage.”
To get whole visibility into agent conduct, Atlassian’s Teamwork Graph collects session information accessbile from anyplace in Jira, the corporate introduced, together with new hooks within the Teamwork Graph CLI that may hyperlink native agent classes on to work in Jira, updating context repeatedly to keep away from agent drift.
In keeping with Atlassian, Jira for Slack, Jira Coding Agent, Jira agent automations, agentic templates, and agent classes in Jira can be found at the moment for paid Jira Cloud clients at no further value. Jira Planner is out there in early entry, and Codex in Jira is coming quickly. DX AI value administration is out there for Atlassian DX clients.
