Use LLM-Based Auto-Approval for Hands-Free Coding Sessions
Reduce workflow interruptions by letting AI classify which actions need human approval.
Activate LLM-based Auto-Approval to allow the AI agent to perform routine, low-risk operations without pausing for manual user confirmation.
The Scenario
You are running an autonomous coding task that involves reading many files and making minor adjustments, and you want to avoid constant confirmation prompts.
Before & after
You have to manually click 'Y/N' or 'Accept' for every single file read or small edit the AI makes, interrupting your focus 10-20 times per task.
Use the /goal command with Auto-Approval; the LLM classifier will silently handle simple tasks and only prompt you for high-risk changes, taking 30 seconds of your attention.
The Prompt
/settings set auto_approval true /goal [TASKS_REQUIRING_MULTIPLE_FILE_READS_AND_MINOR_EDITS]
Auto Approval uses a dedicated LLM classifier to distinguish between routine actions (like reading a file) and sensitive actions (like deleting code), only interrupting the user when necessary.
Source
Blog | Qwen Code Docs"Auto Approval uses an LLM classifier to decide whether human confirmation is needed, so routine operations stop interrupting you."