Summary
Add the ability to configure trusted commands/scripts that can run without approval prompts in Cursor’s integrated terminal.
Problem
Currently, Cursor requires manual approval for all external commands executed from the integrated terminal, including:
- User-created scripts in trusted locations (e.g., ~/bin/)
- Common development tools and utilities
- Custom aliases and functions that call external programs
- Even simple utilities like text-to-speech commands
This creates friction for power users who frequently run trusted scripts and tools.
Proposed Solution
Add a configuration option to specify trusted commands/scripts that can run without approval:
Option 1: Trusted Paths
{
"terminal.integrated.trustedPaths": [
"~/bin/",
"~/.local/bin/",
"/usr/local/bin/"
]
}
Option 2: Trusted Commands
{
"terminal.integrated.trustedCommands": [
"espeak-ng",
"~/bin/speak",
"~/bin/say"
]
}
Option 3: Trusted Scripts with Hash Verification
{
"terminal.integrated.trustedScripts": {
"~/bin/speak": "sha256-hash-here",
"~/bin/say": "sha256-hash-here"
}
}
Benefits
- Reduces workflow friction for power users
- Maintains security for untrusted code
- Allows customization of security boundaries
- Preserves the approval system for unknown commands
Security Considerations
- Only apply to user-specified trusted paths/commands
- Maintain approval for commands from untrusted sources
- Option to require explicit user confirmation for new trusted items
- Clear visual indication when trusted commands are executed
Use Cases
- Development scripts and utilities
- Custom automation tools
- Text-to-speech and accessibility tools
- Common system utilities
Implementation Priority
Medium - This would significantly improve the developer experience for power users while maintaining security.
Additional Context
This feature would be particularly valuable for developers who:
- Use custom scripts for automation
- Work with accessibility tools
- Have established workflows with trusted utilities
- Need to run frequent, safe commands without interruption
The current approval system, while secure, can be overly restrictive for legitimate development workflows.