Skip to content

Permissions

NeuroPilot’s main safety feature is the permissions system. This allows almost fine-grained control over what Neuro is allowed to do and what isn’t allowed.

NeuroPilot actually has 2 main types of permissions: action permissions and non-action permissions.

A constant with action permissions is that they usually have 3 options for their permissions: Off, Copilot, and Autopilot.
If you can’t decide which one to choose, visit this page.

All permissions are set to ‘Off’ by default, except one.

This category of permissions covers actions related to files themselves.
To see the features covered by these permissions, go to this page.

The following permissions belong in this category:

This permission covers actions relating to editing the currently open document.
To see features covered by this permission, go to this page.

The associated permission is found here.

This category covers actions that interact with the local Git repository.
To see features covered by these permissions, visit this page.

The following permissions belong here:

Note: For Git operations to work correctly, certain underlying OS-level and workspace configurations must also be in place.

This category covers actions that can execute commands in a shell. These requires some pre-defining, which you can learn more about here.

The following permissions belong here:

This permission covers actions about getting linting problems from language servers. This can get any linting diagnostic from any extension contributing them. To see features covered by this permission, visit this page.

The associated permission is found here.

Some additional permissions control other interactive elements in the extension.

This special permission controls if Neuro is allowed to ask for cookies. She can also ask for a specific type of cookie.

This permission is set to Copilot by default, and lacks an Autopilot option.

The associated permission is found here


These permissions don’t directly influence action registration (mostly), but still give Neuro more access in some way. These permissions are also disabled by default (where applicable).

  • Allow Unsafe Paths - Whether or not to ignore the built-in path filter.
  • Include/Exclude Patterns - Allows custom definition of safe/unsafe paths. These will still be used to check path validation access even if Allow Unsafe Paths is enabled.
  • Allow Running All Tasks - Registers all tasks as actions, regardless of the task’s prefix in its detail property.