Composition

Base First, Wrapper Second

If the semantic base exists, wrap it.

Prefer:

export const MentionPlugin = toPlatePlugin(BaseMentionPlugin);

over re-declaring the same semantics in createPlatePlugin.

Use Nested Plugins Intentionally

• Put semantic child plugins in the base plugin's plugins array when they are part of the semantic contract.
• Put Plate child plugins in the wrapper when the wrapper owns the React/Plate layer.

Good example:

export const CodeBlockPlugin = toPlatePlugin(BaseCodeBlockPlugin, {
  plugins: [CodeLinePlugin, CodeSyntaxPlugin],
});

Configure, Don't Clone

If you need to adjust a nested child plugin, use configurePlugin instead of copy-pasting a new child plugin definition.

That keeps the ownership boundary obvious and preserves shared behavior.

Extend The Right Surface

• extend Merge config or compute context-aware config.
• extendPlugin Reach into a nested plugin when the parent owns that decision.
• overrideEditor Change editor behavior itself.
• handlers Use for events that actually belong at the plugin boundary, not as a dumping ground for logic that should be a transform or editor override.

Prefer transformProps For React-Only Node Augmentation

When the job is "augment props on already-rendered nodes in the Plate layer", prefer inject.nodeProps.transformProps.

This is especially good when:

• the augmentation is React-only
• hooks are needed
• the semantic base plugin should stay clean
• you do not need to replace the component, only decorate its props

Good fits:

• BlockSelectionPlugin
• NavigationFeedbackPlugin

Do not over-apply this rule. transformProps is for prop augmentation. It is not a blanket replacement for node.component, render, wrapper plugins, or useHooks.

Helpers

Keep helpers inline when they are single-use and context-local.

If you extract a helper:

• make it generic
• keep it context-free when possible
• avoid locking it to SlateEditor by habit

Do not create abstraction sludge just because a callback body is slightly long.