Kotlin

The Kotlin extractor maps tree-sitter Kotlin CST nodes to canonical concepts. The Kotlin grammar declares no field names at all, so every name is recovered from the first identifier-like child: class names from type_identifier, function and variable names from simple_identifier.

Detected concepts

tree-sitter node type	Canonical concept
`class_declaration`	Class
`function_declaration`	Function
`property_declaration`	Variable
`variable_declaration`	Variable
`type_alias`	TypeDef
`import_header`	Import
`package_header`	Namespace
`call_expression`	CallExpression
`assignment`	Assignment
`jump_expression`	Return
`if_expression`	IfBlock
`when_expression`	SwitchBlock
`string_literal`	string_literal

Example

Renaming a function:

old.kt

class Counter { fun total(n: Int): Int { return n + 1 } }

new.kt

class Counter { fun sum(n: Int): Int { return n + 1 } }

Running differens old.kt new.kt:

  ~ renamed function `total` to `sum` in class `Counter`

Example: change the body

Keep the declaration name and change what it does:

class Counter { fun total(n: Int): Int { return n + 1 } }

class Counter { fun total(n: Int): Int { return n + 2 } }

differens body-before.kt body-after.kt

  ~ changed function `total` (`1` → `2`)

The report groups the value edit under its containing declaration. Use --format=json to inspect the underlying action and context, or --format=llm for a compact report with source locations.

Known limitations

The grammar’s missing field names mean label recovery is positional: a declaration with two identifier children can mislabel in exotic syntax.
Extension functions report as functions; the receiver type is part of the structure, not the name.

Semantic Extractors

Tier Adapters

Detected concepts

Example

Example: change the body

Known limitations