[FEATURE] 'Brave Mode' to execute commands without permission prompts
Preflight Checklist
- [x] I have searched existing requests and this feature hasn't been requested yet
- [x] This is a single feature request (not multiple features)
Problem Statement
Currently, when executing commands in Claude Code, the tool prompts for permission before running each potentially impactful action. This can interrupt workflow, especially for advanced users who understand the risks and want to execute a series of commands quickly without repeated confirmations. The manual confirmation slows down automation and batch operations.
Proposed Solution
Introduce an optional 'Brave Mode' (or similar setting/flag) that allows users to execute commands without permission prompts. When enabled, trusted users can run scripts or commands in a streamlined, uninterrupted manner. For example, a CLI flag like --brave or an environment variable could toggle this mode. There should be clear warnings about the risks, and perhaps a config option to restrict this feature to advanced users.
Alternative Solutions
An alternative is to whitelist certain commands, but this requires ongoing maintenance and doesn't help with new or custom actions. Another workaround is to use automation scripts that simulate confirmation, but this is brittle and less secure.
Priority
Critical - Blocking my work
Feature Category
CLI commands and flags
Use Case Example
Scenario:
- I'm running a deployment script that requires several file operations and server commands.
- With current prompts, I have to confirm each action, which is tedious.
- With 'Brave Mode' enabled, I could execute the entire script in one go, trusting my setup and acting quickly.
- This would significantly improve productivity for advanced and power users.
Additional Context
Similar features exist in tools like 'sudo' (with NOPASSWD) or in automation tools that allow skipping prompts for trusted users. Security warnings and documentation should accompany this feature. Mockups are not provided but could be added if needed.
This issue has 3 comments on GitHub. Read the full discussion on GitHub ↗