Qwen Code’s v0.25.1-preview.1 adds a broad set of managed-agent and coding workflow changes, including an experimental Kubernetes runtime and session-based multi-agent collaboration. For developers building agents that need to work across tools and environments, this is a substantial preview, not just a fresh coat of version paint.
Alibaba Qwen Watch analysis
What happened
The preview was published on 9 October. Its release notes describe an experimental Kubernetes CSI runtime with durable worker acknowledgements, background Shell and Monitor runtimes for managed agents, and side tasks in separate web-shell workspaces. The release notes also list session-centric multi-agent collaboration, trajectory search and diagnostic filters, and the ability for hosted turns to use a workspace’s project context.
Other changes include showing captured inputs in hosted approval cards, enforcing per-model concurrency limits on foreground sub-agents, and blocking one MCP server’s rules from authorising another server with a colliding name. The release also includes fixes and documentation work; the long changelog is not a shopping list where every line deserves equal billing.
Our top picks
- Experimental Kubernetes runtime
Managed agents gain a CSI runtime with durable worker acknowledgements. - Background Shell and Monitor
These tools can run in the background for managed-agent workflows. - Session-based collaboration
Agents can collaborate around a session, extending Qwen Code’s multi-agent workflow. - Side tasks in separate workspaces
Web-shell users can keep parallel work apart from the main workspace. - Searchable trajectories
Web-shell users gain trajectory search and diagnostic filters. - More legible approvals
Hosted approval cards can show captured inputs, giving users more context when reviewing a request. - A guard against MCP name collisions
Rules from one MCP server can no longer authorise a different server with a colliding name.
Why it matters
These changes address the plumbing that determines whether coding agents can operate across managed runtimes, multiple sessions and approval steps, rather than merely look capable in a demo. The Kubernetes runtime and background tools are especially relevant to teams testing longer-running, managed workflows.
This is still labelled a preview. The release notes describe what changed, but do not establish how reliably the new runtime or collaboration features perform in production. Operators should test the features against their own setups before handing them anything important. Agents are quite good at acquiring responsibilities; the environment still needs to earn their trust.
Our read
There is enough here to make this a meaningful preview for Qwen Code users working on managed agents. We especially like the combination of new runtime options and more inspectable approval flows. Try the capabilities that match your workflow, but keep the preview label in view and treat the release notes as a map of changes, not a performance guarantee.
What to watch
- Whether the Kubernetes CSI runtime moves beyond experimental status.
- How background runtimes and session-based collaboration behave in real deployments.
- Whether the MCP rule fix and approval-card changes receive further refinement.
Discussion spark: For managed coding agents, which would make you more willing to use them: broader runtime and collaboration features, or clearer approval controls?
Sources and evidence
- Release v0.25.1-preview.1 (9 October 2026, 05:35 UTC)
not affiliated with or endorsed by Qwen