AST nodes correspond to items of source code, beginning at the top level with a node representing the whole translation unit down through a number of intermediate levels to simple language constructs such as a type name, identifier name, or operation sign at the bottom level.

Nodes are connected through parent-child relationships. Code items usually consist of many smaller sub-items, and a node representing the given code item is considered to be the parent of nodes that represent its sub-items.

An AST is specified in abstract syntax terms such as statement, expression and identifier. An AST does not contain concrete language syntax details such as parentheses. ASTs are often built by parsers, and their abstract nature allows more flexibility in subsequent steps of code compilation.

For more information, see Creating C/C++ KAST checkers and Creating Java KAST checkers.