ĀML is structured as a meaning-native compiler and runtime system.
The architecture is designed to transform .aml source files into:
- semantic HTML
- compiler artifacts
- Abstract Meaning Trees
- render accountability logs
- EthicalRenderGate decisions
AML Source (.aml)
↓
Lexer
↓
Parser
↓
Abstract Syntax Tree (AST)
↓
Abstract Meaning Tree (AMT)
↓
Render Evaluator
↓
EthicalRenderGate
↓
HTML Generator
↓
Generated Output
The source layer contains human-readable AML files.
Example:
transmission "061" {
engram eternalFlame {
purpose:
"anchor sovereign identity"
attention_cost:
3
restoration_value:
9
}
}
The source layer is designed to prioritize:
- semantic readability
- low syntactic noise
- clear purpose declarations
- restoration-aware metadata
File:
compiler/lexer.js
The lexer converts AML source text into tokens.
Responsibilities:
- identify keywords
- identify identifiers
- detect braces
- parse strings
- parse numbers
- preserve line and column metadata
Example token:
{
"type": "KEYWORD",
"value": "engram",
"line": 9,
"column": 3
}The lexer is the first deterministic transformation stage.
File:
compiler/parser.js
The parser converts tokens into an Abstract Syntax Tree.
Responsibilities:
- parse blocks
- parse properties
- preserve nested hierarchy
- validate structural relationships
- produce predictable AST output
Example AST node:
{
"type": "Block",
"name": "engram",
"identifier": "eternalFlame",
"children": []
}The parser captures structure before semantic interpretation.
Generated file:
dist/ast.json
The AST represents syntactic