
Volume 344, Issue 1 · 10 articles from 2 issues
Bitbucket Pipelines gets a visual refresh with light-mode logs, step status filters and easier log navigation; configurations are unchanged.
Bitbucket Pipelines gets a visual refresh of the Pipelines list and Pipeline result pages, with no change to configurations, builds, runners, variables or deployments.
New features include a logs-specific light or dark theme toggle, step filtering by status that persists across pipelines, and button-based log navigation with controls moved into the navigation area.
The Pipelines list uses compact colour-coded status labels and a clearer row hierarchy with pipeline names first.
The update uses Atlassian Design System components and tokens, and adds a feedback button.
Atlassian explains its System Health tools that give Cloud admins visibility into site performance under the Shared Responsibility Model.
Atlassian says moving to cloud need not mean losing visibility, and describes System Health for admins on large Cloud sites.
Under its Shared Responsibility Model, Atlassian handles infrastructure and scaling, while customer configuration affects site health, such as growing data, unretired fields and work types, broad filters and queries, and network proxies that slow downloads.
Incident Insights came first, giving a unified view of potential and active incidents across a customer's sites for core apps.
System Health is designed to show what is happening and give admins tools to fix issues themselves.
Atlassian introduces the Context Quality Score, rating enterprise data for AI on completeness, connectedness and relevance.
Atlassian presents the Context Quality Score (CQS), a single metric for how well enterprise data supports AI agents.
It rests on three pillars: completeness (depth of context ingested into the Teamwork Graph from enterprise systems), connectedness (mapped links across seven context types, returned in a permission-aware way) and relevance (whether the model gets curated context without crossing permission boundaries).
Atlassian says its research shows higher quality outputs with significantly fewer tokens per run.
A whitepaper explains how CQS finds gaps in context.
Atlassian argues agent sprawl is an identity and accountability risk and recommends governing AI agents with central policies and visibility.
The post says agent sprawl differs from shadow IT because unapproved actors use an organization's data and permissions to take actions.
It cites Gartner projecting over 150,000 agents at the average Fortune 500 by 2028 with only 13% of organizations feeling they have proper governance, and OutSystems finding 94% worry about AI sprawl.
Consequences include data exposure, excess permissions, unclear ownership, unplanned cost and inconsistent outputs.
Its recommendation is to govern rather than block AI, with central guardrails, visibility into agent usage for cleanup and consolidation, and shared context so agents deliver accurate results.
Atlassian's Jira autumn 2026 release speeds up boards, adds agent visibility and usage reporting via the AMP protocol, and new admin defaults.
Atlassian's autumn release makes company-managed boards load 22% faster, with visible subtasks and epics, saved views and column control.
The work item layout is cleaner, agents can be launched from it, and a new create experience suggests fields for review.
It introduces AMP, the Agentic Multiplayer Protocol, giving each agent its own identity in Jira, plus agent space settings and an agent usage dashboard in the Reports tab.
Admin features help keep configurations consistent across teams, and details are at jirareleases.atlassian.com.
Atlassian makes Code Reviewer, an AI agent for pull requests in Bitbucket Cloud and GitHub, generally available with paid Jira licenses.
Atlassian says AI-generated code has increased review load, with teams relying heavily on AI waiting 71 minutes for human review, up from 32, and median pull request size doubling.
Code Reviewer leaves inline comments flagging bugs, logic errors, security risks and maintainability issues, checks changes against acceptance criteria in linked Jira work items, and summarizes large changes.
A quality filter removes low-value comments, and GitLab support is coming soon.
Internally, Atlassian saw median pull request cycle time drop 31 percent, and a beta Standards feature lets organizations assign coding standards to repositories.
Atlassian and DX data show AI tool adoption above 95% among engineers, but innovation output has not kept pace with the gains.
This Inside Atlassian session draws on DX's Q2 AI Impact Report covering more than 500 organizations.
AI tool adoption among software engineers exceeds 95%, and engineers regularly using AI report saving more than 6 hours a week, with median pull request throughput doubling.
Onboarding has improved, with time for a new engineer to merge their 10th pull request down roughly 57-58% since early 2024.
Yet overall innovation output has barely moved, so the article urges leaders to stop measuring adoption and focus on end-to-end system capacity, connecting organizational context, orchestrating agents across the lifecycle and governing workflows with telemetry.
Atlassian adds step-level permissions and short-lived container tokens to Bitbucket Pipelines and outlines breaking permission changes through 2027.
Atlassian is introducing granular step-level permissions and container tokens in Bitbucket Pipelines to reduce reliance on long-lived credentials stored in pipeline variables.
Each step now gets a short-lived OAuth token, isolated to the build container and repository and valid only while the step runs, with permissions set via auth.system.scopes.
Repository admins can configure branch restrictions to trust pipelines from specific source branches, so release pipelines can push version bumps to main without bot user exemptions.
Breaking changes include a new pipeline:write scope for the Run a pipeline API, a November 30, 2026 change to branch restriction behavior, and the end of broad default step permissions on April 30, 2027.
Atlassian lets customers connect Workday, Entra ID or Google people data to Teamwork Graph for org charts, people cards and better Rovo answers.
Atlassian's Teamwork Graph connects work across Atlassian and connected tools, but titles, teams and reporting lines live separately in HR systems and identity providers.
New people data connectors for Workday, Entra ID and Google bring in who people are, their teams, managers and collaborators.
That enables org charts and richer people cards, and lets Rovo understand which team owns a project, who leads it and which updates come from those closest to the work.
Rovo can then ground answers in more authoritative sources, point users to the right person and personalize responses based on a user's role and team.
Atlassian added AIOps to Jira Service Management with Prevention Center, Incident Command Center and MCP links to Datadog, Sentry and Splunk.
Atlassian says AI-accelerated software delivery means IT operations, SRE and platform teams must quickly understand what happened, why and who can help, without searching fragmented dashboards and chats.
It has introduced AIOps capabilities in Jira Service Management, now part of Service Collection, to support a continuous operational loop across the AI SDLC.
Prevention Center is an AI-native, human-in-the-loop experience that brings change context and risk information together so teams can focus on risky changes before release.
Incident Command Center helps coordinate response and reach root cause faster, while MCP-powered integrations with Datadog, Sentry and Splunk bring observability context into one place.