Atlassian Corporation, a provider of AI-powered collaboration and team productivity software, has announced Code Context, a new capability within the Atlassian Teamwork Graph that provides developers, Rovo, and coding agents with secure access to source code across large-scale, multi-repository codebases. The move positions Atlassian as a context provider for the growing ecosystem of AI coding tools — including third-party agents like Cursor, Claude Code, and Codex — rather than building a standalone code assistant, distinguishing its approach from competitors focused primarily on in-editor code generation.

As organizations increasingly adopt coding agents, the effectiveness of these tools depends on more than model intelligence. It relies on the context available to them before they take action. Coding agents can write, refactor, and review code, but producing quality output requires a deep understanding of the system of work surrounding any given task, along with permission-aware access to the context it holds. Organizational knowledge — from architectural decisions and product strategy to implementation discussions, prior tradeoffs, and the conversations that explain why systems function the way they do — is critical to this process. This gap between model capability and contextual awareness has been a recurring challenge for enterprises deploying AI coding tools at scale, where codebases span hundreds or thousands of repositories.

The Atlassian Teamwork Graph now incorporates Code Context, creating a unified context layer where large-scale codebase understanding and organizational knowledge converge. This enables agents to reason across both code and work efficiently, producing more precise output with fewer tokens.

From Code Search to Codebase Understanding

Code Context generates a secure, queryable representation of connected codebases and makes that knowledge accessible across development workflows — from IDEs and terminals to AI coding applications, Jira, Rovo Chat, and the broader Atlassian ecosystem. Developers and agents can leverage exact search, natural-language queries, and semantic retrieval to locate relevant source code across multiple repositories.

Through the Teamwork Graph CLI, coding agents such as Cursor, Claude Code, and Codex can ground their work in relevant information before planning, generating, or reviewing code. Without Code Context, agents are typically limited to the code available locally on a developer's machine, potentially missing implementation details, dependencies, or usage patterns that reside elsewhere. This can create risk when changes fail to account for the system as a whole.

Code Context also allows agents to combine code with related signals from Jira work items, Confluence pages, Loom videos, and third-party sources through more than 50 connectors. In Atlassian's internal benchmarks, agents enriched by the Teamwork Graph delivered 44% more accurate results while using 48% fewer tokens compared to agents operating without that context.

Mark Walz, Chief Technology Officer of SpotOn, said: "When an agent starts in the wrong place, everything slows down and costs more. Developers end up explaining where to look, why the code works the way it does, and what else it touches. Code Context puts all of that in front of the agent from the start, so it can focus on getting the work done."

The AI Context Engine for Software Development

Code Context expands Atlassian's vision for an AI-native software development lifecycle in which Jira anchors the work, Confluence holds the knowledge, pull requests capture implementation history, and the Teamwork Graph connects the surrounding context. Large-scale, complex codebases are now part of that connected system. This integration leverages Atlassian's existing footprint across enterprise software teams — Jira and Confluence are already deeply embedded in many organizations' development workflows — giving Code Context a natural distribution channel that standalone code intelligence tools may lack.

The Teamwork Graph continuously cross-references, interlinks, infers, indexes, and pre-calculates relationships among billions of objects, constructing a unified map of how everything relates. This provides agents a path from hypothesis to verification, using relevant context to validate their reasoning before taking action. Rather than manually examining repositories or relying on tribal knowledge, developers can ask Rovo how a system works and receive answers grounded in connected source code and organizational context.

For engineering teams adopting coding agents, Code Context reduces the time spent manually assembling context. Agents gain a more direct path to the information they need before planning, generating, or reviewing code. Developers retain accountability for understanding changes and determining what ships.

Sanchan Saxena, Senior Vice President and Head of Product for Teamwork Collection at Atlassian, said: "The organizations seeing the most success with AI aren't just spending the most; they're the ones giving it the best context with the right access. Frontier intelligence gives teams speed. Context points that speed in the right, most impactful direction. We're finding that real acceleration in software delivery comes from combining both."

Code Context is designed with security and governance at its core. Once enabled, results are scoped to what each user or authorized agent is permitted to see. The objective is not to grant agents unlimited access, but to provide the right level of access through a secure, governed system in which developers remain in control. The emphasis on permission-aware access addresses a key concern for enterprises evaluating AI coding tools, where unrestricted codebase exposure has been a barrier to adoption in regulated industries.