Atlassian unveiled Code Context, a new feature embedded in its Teamwork Graph, designed to equip AI coding agents with comprehensive understanding beyond isolated code, combining cross-repository insights with organizational context to accelerate and secure development workflows.
- Code Context combines code and organizational data for richer AI agent insight
- Enables cross-repository queries and natural language code searches
- Improves accuracy by 44% and reduces token use by 48% in internal testing
What happened
Atlassian introduced Code Context, a new capability built into its Teamwork Graph platform that offers AI coding agents a more holistic view of software projects. By indexing entire codebases alongside relevant organizational materials such as Jira tickets, Confluence documentation, and communication tools like Slack, Code Context facilitates richer, more informed interactions for coding agents. Developers can access this contextual information directly within popular IDEs, terminals, and AI coding environments.
The system works by linking lexical and semantic code search functionalities with the broader organizational knowledge base, allowing agents to move beyond isolated code snippets to understand architectural decisions, product strategies, and cross-team dependencies. Early internal benchmarks reported up to a 44% increase in accuracy and a 48% reduction in token consumption when agents leverage this combined data, underscoring significant efficiency gains.
Why it matters
AI coding agents traditionally face challenges similar to new developers starting on a codebase—they lack sufficient context to grasp complex systems, project history, and collaboration dynamics. This gap often causes slower output, greater risk of introducing errors, and increased dependency on human intervention to clarify where and why code changes should occur.
Code Context closes this gap by providing secure, permission-controlled access to a comprehensive graph of code and related organizational signals, enabling agents to make targeted, context-aware recommendations. This reduces costly misunderstandings and streamlines development workflows, allowing human developers to focus more on decision-making and critical review rather than basic navigation and explanation.
What to watch next
Organizations interested in deploying Code Context need to enable code indexing through Atlassian’s admin settings. This feature integrates seamlessly with existing Atlassian tools and supports prominent AI coding agents like Cursor, Claude Code, and Codex, further broadening its adoption potential. Watching how teams integrate and measure improvements in real-world environments will be crucial to understanding the feature’s practical impact.
Future developments may include expanded connector support for additional third-party collaboration tools and enhanced conversational AI capabilities such as those demonstrated by Rovo Chat, which can answer system-level queries by referencing this rich contextual data. Monitoring Atlassian’s roadmap and community feedback will reveal how Code Context evolves and influences AI-driven software development practices.