Deploy an AI Agent Team for Parallel Coding Workflows
Orchestrate multiple specialized AI agents to handle complex, multi-part coding projects.
Use the 'Agent Team' feature to spawn persistent, collaborative agents that share task lists and message each other to consolidate large projects into a single report.
The Scenario
You are working on a complex codebase that requires simultaneous refactoring of logic and the creation of new unit tests, tasks too large for a single linear chat.
Before & after
You would manually open multiple chat windows or use /fork to run independent agents, then spend 30-45 minutes manually reading and merging their different code outputs and suggestions.
Enable the experimental Agent Team feature to have Qwen orchestrate a lead agent and specialized teammates who message each other and merge their results into one report. This takes about 2-3 minutes to set up.
The Prompt
Enable Agent Team mode and run: Create an Agent Team to refactor this repository. Assign one teammate to write unit tests for [MODULE_NAME] and another to optimize the database queries in [FILE_PATH]. Provide a consolidated report when finished.
The Agent Team mode allows for persistent teammates that don't disappear after one task, enabling a structured multi-agent hierarchy within a single session.
Source
Qwen Code Blog: AI Coding Updates and Guides | Qwen Code Docs"Agent Team mode lets the model create named teams, spawn multiple teammates to work in parallel, exchange messages, share task lists."