claude-code - 💡(How to fix) Fix Agent launches background sub-agents without user approval, wasting tokens [1 participants]

TL;DR

Modify Claude Code to wait for explicit user approval before launching background agents for token-intensive tasks like SVG icon generation.

Guidance

• Review the code responsible for launching background agents and ensure it checks for user approval before proceeding, especially for tasks marked as token-intensive.
• Verify that the run_in_background: true parameter is only set after receiving explicit user confirmation for the specified task and categories.
• Implement a mechanism to prompt the user for approval before launching any background agents, adhering to the principle of confirming before taking expensive or irreversible actions.
• Consider adding a feature to estimate the token cost of a task before launching it, allowing the user to make an informed decision about proceeding.

Example

No specific code example can be provided without the actual codebase, but the fix should involve adding a conditional check for user approval before the line that launches the background agents.

Notes

The solution assumes that the issue lies in the logic of launching background agents without user approval. The actual implementation details may vary based on the specific code structure and requirements of the Claude Code project.

Recommendation

Apply a workaround by modifying the code to require explicit user approval for launching background agents, especially for token-intensive tasks, to prevent unnecessary token consumption and improve user experience.