--- ---

Documentation Style Guide

DocOps Lab documentation style guide and AsciiDoc syntax conventions.

DocOps Lab is an AsciiDoc shop. With a few exceptions, all technical documentation is sourced in AsciiDoc format using a particular (standards-compliant) syntax style.

Structured/reference documentation is typically stored in YAML-formatted files, often with AsciiDoc-formatted text blocks.

Some documentation in DocOps Lab projects is written in Markdown format, such as documents intended for AI consumption (such as for agent orientation/instruction or for RAG retrieval).

Automated Style Enforcement

DocOps Lab projects using the docopslab-dev tool automatically enforce documentation style guidelines. This is done using Vale, a prose and source-syntax linter.

To check documentation style:

Check prose for style issues
bundle exec rake labdev:lint:text
Check for AsciiDoc markup syntax issues
bundle exec rake labdev:lint:adoc
Check both syntax markup and prose
bundle exec rake labdev:lint:docs

See DocOps Lab Dev-tooling Setup for more on the docopslab-dev tool. For Vale configuration details, see Vale Configuration and Usage.

DocOps Lab maintains a general-audience style guide in the AYL DocStack project repository and website. That guide is reproduced here.

General AsciiDoc Syntax Guidelines

DocOps Lab documentation largely follows the conventions outlined in the Recommended Practices andWriter’s Guide documents maintained by the Asciidoctor project.

Reinforcements and exceptions:

  • Use .adoc extensions execpt for Liquid templates used to render AsciiDoc files, which use .asciidoc.

  • Use one sentence per line formatting.

    • Let hard-returns signal spaces between sentences.

    • Also do this for major colon- or semicolon-delimited sentences.

  • Use ATX-style titles and section headings.

  • For DRYness, use attributes for common URLs and paths (see Attribute Formatting).

DocOps Lab Specific Syntax Guidelines

Inline Syntax

Inline Semantics

The main purpose of inline semantics is to provide a clear indication of the role of the text to the reader — including artificial readers.

We can convey semantics by way of:

  • declaration by element, role, or class

  • text style based on declaration

  • browser effects based on declaration and additional data

We use the following inline semantic coding in DocOps Lab publications.

Syntax Preferences

Use inline semantics liberally, even if you only insert the heavier syntax on a second or third pass.

Formatting with simple *, _, and ` characters on first drafting makes lots of sense — or even missing some of these altogether until the second pass.

But before you merge new text documents into your codebase, add role-based inline semantics wherever they are supported.

Let the reader know and make use of special text, most importantly any verbatim inline text.

Even if you are not ready to add such fine-grained tests to your pipeline, consider the value of having all your commands for a given runtime app labeled ahead of time (such as .app-ruby), and the advantage to the reader, as well.

Block Syntax

Block Semantics

Use semantic indicators deliberately.

The more you assert about a block of text you are writing, the better the placement and content of that block will be.

Semantic assertions reside in the source markup, which may convey means of interpreting that same data visually in the output, as an indication to the reader.

For instance, warning admonitions should only deliver warning content, and the user should clearly see that a warning is interrupting the flow of the content in which it appears.

[WARNING]
====
Avoid misusing or overusing admonition blocks.
====

Semantic notations in our source remind us to treat the content properly.

[WARNING]
====
Avoid misusing or overusing admonition blocks.
This will be hypocritically violated throughout this guide.
====

True as it may be, the second sentence in that admonition should be removed from the block. It can either be its own block, or it can be allowed to fade into the surrounding content.

Sometimes the entire admonition may end up deserving this treatment.

Use Delimited Blocks

Generally, use explicit boundary lines to wrap significant blocks, rather than relying on other syntax cues to establish the “type” of block is intended. These lines are called linewise delimiters.

For example, use the following syntax to wrap the contents of an admonition block:

Example 1. Example admonition block syntax with linewise delimiter
[NOTE]
====
The content of an admonition block should be sandwiched between `====` lines.
Use one-sentence-per-line even in admonitions.
====

The standard linewise delimiters for various AsciiDoc blocks are as follows:

====

For admonitions and examples
----

For code listing (verbatim) blocks
....

For literal (verbatim) blocks
****

For sidebar blocks
|===

For tables
____

For quote blocks
++++

For raw/passthrough blocks
--

For open blocks

For code listings, literals, or really any block that might contain text that could be confused with the delimiter, vary the length by using a greater number of delimiter characters on the outer block.

Example “example” block containing an admonition block
[example]
========
[NOTE]
====
This is an example block containing an admonition block.
====
========

Exception: Brief admonitions

Some blocks do not require delimiters. In cases of repeated, nearly identical blocks, containing just one line of content, you can use the single-line syntax where it is available.

Example single-line admonition block syntax
NOTE: This is a single-line admonition block.
Exception to this exception

We do not recommend the same-line syntax for admonition blocks other than NOTE and TIP. For IMPORTANT, CAUTION, and WARNING, use at least the 2-line syntax, if not explicit delimiters.

[IMPORTANT]
This is a critical notice, but it's not warning you of danger.

Exception: Single-line terminal commands

Another common case is 1-line terminal commands, for which this guide recommends using a literal block with a prompt role added.

[.prompt]
 echo "Hello, world!"

The single preceding space notation affirms the use of a literal block for any consecutive lines of content preceded by a single space. For multi-line terminal commands/output, use the …​. syntax to distinguish the block.

Exception to the exceptions

Whenever additional options must be set for a block, such as a title or role, use the linewise delimiter syntax — even in one-liner cases.

[.prompt,subs="+attributes"]
....
echo "Hello, {what}!"
....

Example Blocks

Use example blocks liberally. If something fits the description of being an example — especially if the words “example” or “sample” are used in the title, caption, or surrounding text referring to a given block of anything…​ then wrap it in an example block.

Instances of the following block types may commonly be instances of examples, and just as commonly they may not be.

  • figures (diagrams, illustrations, screenshots)

  • tables

  • code listings

  • literal blocks (sample prompts, logs, etc)

  • rich-text snippets (rendered results, a user story, etc)

Whenever any such instances are examples, prepend and append them with example blocks, and prefer to title them at the exampple-block level rather than the inner-content level.

Example of a code block treated as an example
:example-caption: Example

.require statement in Ruby
====
[source,ruby]
----
require 'jekyll'
----
====

Special Syntax

Attributes

Attribute Formatting

AsciiDoc attributes are often used to store reusable matter. In certain contexts, attributes should follow a formatting convention that makes them easier to name and recall.

Boolean Attributes

Use toggles to set or conditionalize states such as:

  • intended audience type or role

    • audience-agent

    • audience-beginner

    • ``

  • target platform or format

    • env-github

    • site-gen-jekyll

    • backend-pdf

These kinds of attributes are passed depending on how the AsciiDoc is converted. Platform and format indicators tend to get argued by the converter at runtime.

But you can also look check for statuses that might be set in previous files depending on the use-case of the output.

Testing for existence of a target platform
ifdef::audience-level-beginner[]
As a beginner, you will see extra content in parts of this guide.

If you are an expert, skip to the <<expert-guide>>.
endif::[]
Testing for non-existence of a target audience type.
ifndef::audience-agent[]
This content is _not_ to appear in docs generated for AI agents.
endif::[]

It is generally advised to create two versions of any such indicator that may need to be resolve a variable placeholder later.

Setting open-ended key and boolean simultaneously
:audience-level: beginner
:audience-level-beginner: true

Later we can reference the {audience-level}, which might be overwritten by an attribute passed at runtime.

URL Attributes

Format URL-storing attributes like so:

:syntax_area_descriptive-slug_form:

Where:

  • syntax_ is one of

    • href_ (external)

    • xref_ (local)

    • none (skip it — presumed to be a straight URL)

  • area_ is a component or category like docs_ or pages_, mainly to ensure unique slugs across divisions

  • form is the way the resource is presented:

    • link (includes linked text and the URL)

    • url (just the URL)

Examples
:docopslab_hub_url: https://github.com/DocOps
:href_docopslab_aylstack_url: {docopslab_hub_url}/aylstack/
:href_docopslab_aylstack_link: link:{href_docopslab_aylstack_url}[AYL DocStack]

Vale Configuration and Usage

Vale configuration and styles are managed in coordination with the link:`docopslab-dev` gem.

Our implementation of Vale allows for local project overrides while maintaining a centralized database of styles.

Linting for documentation quality and consistency, both AsciiDoc markup syntax and prose quality/correctness.

This tool provides a custom styles package and a modified configuration system, enabling multi-file merging.

Base config

.config/.vendor/docopslab/vale.ini (from source)

Project config

.config/vale.local.ini (inherits via BasedOnStyles)

Ephemeral config

.config/vale.ini (merged from base and target)

Sync command

bundle exec rake labdev:sync:vale

Consumer Mode (Other Projects)

For all other projects, the gem works in a standard package consumption mode:

  • The project’s vale.ini should list all desired packages, including a URL to the stable, published DocOpsLabStyles.zip.

  • The labdev:sync:styles task simply runs vale sync in the proper context, downloading all listed packages into a local .vale/styles directory.

The labdev:sync:vale task updates both the base config and the styles package.

The .config/vale.ini for consumer projects (based on the gem’s template) should look like this:

# CONSUMER MODE CONFIG

StylesPath = .vale/styles

# List all packages, including the URL to the central DocOpsLabStyles package.
# TODO: Update with the real URL.
Packages = RedHat, proselint, write-good, https://example.com/path/to/DocOpsLabStyles.zip

[*.adoc]
BasedOnStyles = RedHat, DocOpsLab-Authoring, DocOpsLab-AsciiDoc

This dual-mode system provides a robust workflow for both developing and consuming the centralized Vale styles.

For full Vale configuration settings (“keys”) reference, see the Vale documentation.
For information on managing DocOps Lab’s Vale styles, see the docopslab-dev gem README.