A lookup list for every core AEON value form.

Escapes and quoting

Strings are best for compact text. Use escapes when the delimiter or a control character must live inside the value.

The spec distinguishes ordinary quoted strings from raw backtick strings. Quoted strings support Unicode escapes, but malformed escape sequences fail closed.

Use trimtick/prose when the text is naturally multiline; it keeps source readable without a wall of escape sequences.

quote:string = "She said \"AEON\"."
path:string = "C:\\notes\\today"
line:string = "first\nsecond"

Marker width

Trimticks are multiline strings with indentation trimming. The number of leading > markers gives tools a stable tab width while normalizing indentation.

For example, >>>> treats a tab as four spaces before common indentation is trimmed. The marker must be contiguous, though spaces before the backtick opener are allowed.

body:prose = >>>>`
	Indented with a tab
	and normalized as four spaces.
`

Finite numbers

Numbers cover finite integer and decimal forms. Numeric width and precision are not Core-level semantics.

Underscores are valid only between digits. Canonicalization removes underscore separators and normalizes leading-dot decimals to an explicit zero.

count:number = 42
ratio:number = 3.14
large:number = 1e3
readable:number = 1_000

Explicit edge case

NaN is not an ordinary number. Accepted surface forms are exactly NaN and -NaN; +NaN and lowercase aliases are invalid.

Explicit bounds

Infinity and -Infinity are useful for open-ended bounds, but remain distinct from finite numbers. Strict JSON-profile output must fail closed rather than silently rewriting them.

Narrow truth

Boolean values are lowercase true and false. TRUE is not a boolean token, and quoted "true" is still a string.

Words carry intent

On/off often reads like operational state. Yes/no often reads like a human answer. AES preserves the lexical value, while JSON materialization maps yes/on to true and no/off to false.

Absence has reasons

Use reserved null reasons when they fit, and custom quoted reasons when absence has domain-specific meaning.

Custom reasons must decode to a non-empty, non-whitespace string and must not collide with reserved sentinel names.

middleName:null = !none
phone:null = !notSet
fax:null = !notApplicable
oldId:null = !tombstone
review:null = !"awaiting manual review"

Prefix plus datatype

The # prefix identifies the literal family. The surfaced value is payload-only, while raw preserves the source token. Underscores are visual spacing only and valid only between hex digits.

Base-aware digits

The % prefix carries the digits, while bracket metadata declares the base. Core accepts bases 2 through 64, but lexical acceptance is still separate from base-specific digit validity.

Encoded payloads

The $ prefix makes the payload shape visible. Core accepts standard and URL-safe base64 alphabets; canonical encoding prefers URL-safe output and strips trailing padding.

Calendar only

A date literal is a calendar date without a time of day. Core recognition includes intrinsic calendar validation, including leap-year handling.

Time of day

A time may be local, UTC, or offset-bearing. Core allows reduced-precision forms such as 09: and 09:30Z, but rejects out-of-range clock values.

Instant and zone authority

A datetime can express an instant. ZRUT adds a named local time authority when the zone itself is part of the meaning. Uppercase Z is the UTC marker; lowercase z is rejected as a temporal literal marker.

published:datetime = 2026-05-05T09:30:00Z
localPublication:zrut = 2026-05-05T09:30:00Z&Australia/Melbourne

Separator grammar

Use sep[...] when the separator character is part of the value grammar. Separator specs must name exactly one allowed character, though whitespace may appear around it inside brackets.

Quoted "..." segments inside separator payloads use ordinary AEON string lexical rules, so content like labels or spaces can be carried without becoming raw separator syntax.

Kadot is a reserved subset name for dot-separated numeral lists, but Core only enforces the separator literal family; stricter kadot shape validation belongs to schema/profile layers.

dimensions:sep[x] = ^300x250
triple:sep[x][y][z] = ^100x200y300z
row:dsv = ^"Name"."Address"."phone"."email"
ip:kadot = ^198.0.126.255

Named structure

Objects create member paths and are the natural shape for records, configuration, and grouped claims. Reserved aliases such as obj, o, and envelope perform the same Core compatibility check.

Order and duplicates

Lists preserve order and duplicates. Generic args are optional syntax and are not enforced by Core semantic typing rules.

Position is meaning

Use tuples when each position has a role and the length is part of the contract. AEON Core preserves tuple/list distinction in AST and AES; they must not collapse into one sequence kind.

Tagged tree values

Nodes carry tagged children and attributes. Empty-node shorthand <tag> is equivalent to <tag()>, and node children use indexed canonical paths rather than becoming top-level bindings.

Copy intent

Clone references say that this value should use the value at another path. Missing, forward, and self references are illegal in Core, before AEOS gets involved.

Pointer intent

Pointer references preserve alias/pointer intent in AST and AES. JSON interoperability may lose that identity distinction once references are materialized.