node package manager


cod Build Status Module Version Dependency Status

An unassuming documentation format that works with any language.

  Text can go anywhere.
     Whitespace is preserved.
  @number 42
  @string Hello, cod
    @property yay
    Nested text.
  @list A
  @list B
  @list C
  "Example": {
    "!text": "Text can go anywhere.\n   Whitespace is preserved.",
    "flag": true,
    "number": 42,
    "string": "Hello, cod",
    "nested": {
      "!text": "Nested text.",
      "property": "yay"
    "list": ["A", "B", "C"]

stupid by design

cod is not a documentation generator.

It is a documentation format that outputs nearly "1-to-1" JSON.

cod does zero code analysis.

cod doesn't know what @class, @return, @param or @any @other tag means.

cod does not generate HTML, PDFs, or any traditionally human-readable documentation; that part is left up to you.

As such, cod is not a standalone replacement for tools like doxygen or Sphinx, but it functions as the first step in a more flexible doc-generation process for those who need finer control.

cod may not be smart, but it's not stubborn, either.

You write your docs in cod's format, and it faithfully outputs JSON. That's it.

Use whatever tags you need.

Anything that isn't a @tag is text.

Text sections are left untouched. You can process it as Markdown later. Or HTML. Or just keep it as plain text.

Once you have the JSON, use it however you like:

  • Utilize existing templates and styles.
  • Build an app that can consume multiple versions of your API docs.
  • Easily compare specific versions of your API at the structural level.

cod is naturally language-agnostic; all it needs to know is the pattern you use to denote a doc-block (i.e. /** and */).


Tags without values are treated as boolean flags:

  "flag": true

There are also numbers and strings, and you can explicitly set a flag to false:

@number 42
@string Hello there.
@boolean false
  "number": 42,
  "string": "Hello there.",
  "boolean": false

Structure is designated by indentation.

  @nested value
  "root": {
    "nested": "value"

Whitespace after indentation is preserved in text blocks.

  This is some example text.
  It can handle multiple lines.
    Indentation is preserved.
  "example": {
    "!text": "This is some example text.\n\nIt can handle multiple lines.\n  Indentation is preserved."

Specifying a @tag more than once will turn it into a list of values.

@list A
@list B
@list C
  "list": [

@tags:like:this are equivalent to nested tags.

@root:inline:nested value
  "root": {
    "inline": {
      "nested": "value"

Values of tags that have nested properties or text bodies are stored as @complexProperty["!value"].

@simpleTag 100
@complexTag This will be stored as example["!value"]
  This allows for nested text and tags.
  "simpleTag": 100,
  "complexTag": {
    "!value": "This will be stored as example[\"!value\"]",
    "!text": "This allows for nested text and other properties.",
    "likeThis": true


cod *.js # or *.go or *.c or ...
cod -b '###*' -e '###' *.coffee
cod -b "'''*" -e "'''" *.py
cod -b '{-*' -e '-}' *.hs
cod -b '--[[*' -e ']]' *.lua
cat *.js | cod  > api.json
cod --help
     .-*'``           ``-.__.-`:
  .'o   ))` ` ` ` ` ` `_`.---._:
   `-'.._,,____...--*"` `"     
cod: An unassuming documentation generator.
  cod [-b <doc-begin> -e <doc-end>] [-o <output-file>] [-u] [<input-file>...]
  cod -h | --help | --version
  -b <doc-begin>    String that marks the start of a doc-block [default: /**]
  -e <doc-end>      String that marks the end of a doc-block [default: */]
  -o <output-file>  Output file [default: STDOUT]
  -u --ugly         Output non-pretty JSON.
  -v --version      Show version.
  -h --help         Show this screen.
  <input-file>...   File(s) containing docs. If none, cod reads from STDIN.


See gulp-cod.


Create an issue if you make a Grunt plugin for cod, and I'll list it here.


#### cod([[text,] options])

If text is supplied, cod will parse it and return a plain JS object that contains your doc structure.

Otherwise, cod will return a Transform stream into which your source can be piped. cod will buffer the stream until completion, after which it will output the stringified JSON of your doc's structure.

text (String | Buffer)

Text containing cod-style documentation. Probably source code.

options (Object)

docBegin (String) default: "/**"

String that marks the start of a doc-block

docEnd (String) default: "*/"

String that marks the end of a doc-block

pretty (boolean) default: true

Format the JSON output with JSON.stringify(doc, null, 2) Only applicable in stream mode when text is not supplied

var cod = require('cod');
var doc;
doc = cod([
  'Hello, cod.',
  '@answer 42',
// Output: 
// { '!text': 'Hello, cod.', 'answer': 42 } 
var fs = require('fs'),
    cod = require('cod');
// ###* 
// Hello, cod. 
// @answer 42 
// ### 
fs.createReadStream(__dirname + '/')
    docBegin: '###*',
    docEnd: '###',
    pretty: false
// Output: 
// {"!text":"Hello, cod.","answer":42} 


npm install -g cod