Delegate Complex Coding Tasks to Autonomous Agents with /goal
Use the /goal command to let AI plan and execute multi-step coding tasks.
Leverage the /goal command to trigger autonomous coding agents that break down complex requests, execute them, and verify the output.
The Scenario
You have a complex feature to implement or a large batch of bugs to fix and want the AI to handle the cognitive load of planning and execution.
Before & after
Developers manually break tasks into sub-tasks, execute them one by one, and verify each result. This process usually takes 45–60 minutes.
Run /goal [TASK] and the agent will use parallel sub-agents and a 'judge model' to complete the job. This takes 2–Relies on autonomous execution in 5 minutes.
The Prompt
/goal Create a comprehensive test suite for [DESCRIBE_MODULE] and fix any bugs found during execution.
The v0.16.0 update introduced /goal for autonomous coding, paired with an independent judge model to detect impossible goals and ensure task quality.
Source
Qwen Code Blog: AI Coding Updates and Guides | Qwen Code Docs"Qwen Code v0.16.0 follows suit—plus an independent judge model and impossible-goal detection."