The friendly data format for human & machine

Release Status Prerelease (Please help by reviewing the specs and submitting issues!)

The Best of Both Worlds

Today's data formats present us with a dilemma: Use text based formats that are bloated and slow, or use binary formats that humans can't read. Wouldn't it be nice to have the benefits of both and the drawbacks of neither?


It's a hostile world!

Did you know that many security breaches happen as a result of slightly different behaviors in parsers? A rounding or replacing Python parser at the access-control level, mixed with a truncating Go parser at the database level, could be exploited to poison your data and elevate privileges!

The Concise Encoding spec is built with security in mind, ensuring consistent, reliable behavior among all codecs to keep your data and system safe.


Green computing is the future.

Our connected world demands more and more data each day, and that data comes at a cost. Data centers currently consume 1% of the world's energy, and the trend is for 3-8% within the next decade.

What if you could cut your data costs significantly by swiching JSON/XML/HTML for a more efficient format?

What if you could cut your data costs AND keep the human viewability/editability of your data?

Save CPU and bandwidth

Concise Encoding has twin 1:1 compatible formats. Store and transmit your data in the smaller, more efficient binary format, and only convert to/from text in the rare cases where a human needs to read or edit.


Tired of shoehorning your data into strings?

We've been conditioned over two decades into believing that data types other than numbers and strings must be "string encoded". This is nonsense! Our data models are type-rich, and so our data formats should support them!

Concise Encoding supports the most important types natively:


Because nobody's perfect.

No specification is impervious to errors-in-specification, or the need to change to meet a new reality. Changing a specification leads to all sorts of incompatibilities and mismatched implementations UNLESS you version both the spec AND the documents.

In Concise Encoding, both specs and documents are versioned, so you can rest secure in the knowledge that a CE decoder will properly handle your CE document forever.

Plug and Play

Extra steps suck.

Concise Encoding can be used with or without a schema. There are no extra compilation phases, and no complicated definition files or code generation required to use it. Just import the library and go!


Concise Text Encoding documents that can be transparently 1:1 converted to/from Concise Binary Encoding.


    "boolean"       = true
    "binary int"    = -0b10001011
    "octal int"     = 0o644
    "decimal int"   = -10000000
    "hex int"       = 0xfffe0001
    "very long int" = 100000000000000000000000000000000000009
    "decimal float" = -14.125
    "hex float"     = 0x5.1ec4p+20
    "very long flt" = 4.957234990634579394723460546348e+100000
    "not-a-number"  = nan
    "infinity"      = inf
    "neg infinity"  = -inf


    "string" = "Strings support escape sequences: \n \t \27f"
    "url"    = @""
    "email"  = @""

Other Basic Types

    "uuid"      = f1ce4567-e89b-12d3-a456-426655440000
    "date"      = 2019-07-01
    "time"      = 18:04:00.940231541/E/Prague
    "timestamp" = 2010-07-15/13:28:15.415942344/Z
    "na"        = na:"database is offline"
    "nil"       = nil
    "media"     = |application/x-sh 23 21 2f 62 69 6e 2f 73 68 0a 0a
                  65 63 68 6f 20 68 65 6c 6c 6f 20 77 6f 72 6c 64 0a|


    "list"          = [1 2.5 "a string"]
    "map"           = {"one"=1 2="two" "today"=2020-09-10}
    "bytes"         = |u8x 01 ff de ad be ef|
    "int16 array"   = |i16 7374 17466 -9957|
    "uint16 hex"    = |u16x 91fe 443a 9c15|
    "float32 array" = |f32 1.5e10 -8.31e-12|


    "main-view" = <View,
        <Image "src"=@"img/avatar-image.jpg">
        <Text "id"="HelloText",
            Hello! Please choose a name!
        // OnChange contains code which might have problematic characters.
        // Use verbatim sequences (\.IDENTIFIER ... IDENTIFIER) to handle this.
        <TextInput "id"="NameInput" "style"={"height"=40 "color"="gray"} "OnChange"="\.@@
            NameInput.Parent.InsertRawAfter(NameInput, '<Image "src"=@"img/check.svg">')
            HelloText.SetText("Hello, " + NameInput.Text + "!")
            Name me!


    // Entire map will be referenced later as $id1
    "marked object" = &id1:{
        "recursive" = $id1
    "ref1" = $id1
    "ref2" = $id1

    // Reference pointing to part of another document.
    "outside ref" = $@""


    // Marked base resource identifiers used for concatenation.
    "resources" = [

    // Map-encoded relationships (the map is the subject)
    $people:"homer_simpson" = {

        /* $mp refers to @"""
         * $mp:"wife" concatenates to @""
        $mp:"wife" = $people:"marge_simpson"

        // Multiple relationship objects
        $mp:"regrets" = [

    "relationship statements" = [
        &marge_birthday:($people:"marge_simpson" $mp:"birthday" 1956-10-01)
        &forgotten_birthday:($people:"homer_simpson" $mp:"forgot" $marge_birthday)
        &firing:($people:"montgomery_burns" $mp:"fired" $people:"homer_simpson")

        // Multiple relationship subjects
        ([$firing $forgotten_birthday] $mp:"contribute" $mo:"marital_strife")


    /* Given: Actual type and value of "eggshell" and
     *        "navy-blue" have been defined in a schema
    "wall color" = #eggshell
    "door color" = #navy-blue

Custom Types

    // Custom types are user-defined, with user-supplied codecs.
    "custom text"   = |ct cplx(2.94+3i)|
    "custom binary" = |cb 04 f6 28 3c 40 00 00 40 40|

See the Resources page or Github page for more information.

Support Open Source Development!