Build process customization

Read the Docs has a well-defined build process that works for many projects. We also allow customization of builds in two ways:

Customize our standard build process: Keep using the default commands for MkDocs or Sphinx, but extend or override the ones you need.
Define a build process from scratch: This option gives you full control over your build. Read the Docs supports any tool that generates HTML.

Extend or override the build process

In the normal build process, the pre-defined jobs checkout, system_dependencies, and upload are executed. If you define a sphinx or mkdocs configuration, the create_environment, install, and build jobs will use the default commands for the selected tool. If no tool configuration is specified, these jobs won’t execute anything by default.

The jobs where users can customize our default build process are:

Step	Customizable jobs
Checkout	`post_checkout`
System dependencies	`pre_system_dependencies`, `post_system_dependencies`
Create environment	`pre_create_environment`, `create_environment`, `post_create_environment`
Install	`pre_install`, `install`, `post_install`
Build	`pre_build`, `build`, `post_build`
Upload	No customizable jobs currently

Note

Any other pre-defined jobs (checkout, system_dependencies, upload) cannot be overridden or skipped.

These jobs are defined using the configuration file with the build.jobs key. This example configuration defines commands to be executed before installing and after the build has finished, and also overrides the default build command for the htmlzip format, while keeping the default commands for the html and pdf formats:

.readthedocs.yaml

version: 2
formats: [htmlzip, pdf]
sphinx:
   configuration: docs/conf.py
python:
   install:
     - requirements: docs/requirements.txt
build:
  os: "ubuntu-22.04"
  tools:
    python: "3.10"
  jobs:
    pre_install:
      - bash ./scripts/pre_install.sh
    build:
      # The default commands for generating the HTML and pdf formats will still run.
      htmlzip:
        - echo "Override default build command for htmlzip format"
        - mkdir -p $READTHEDOCS_OUTPUT/htmlzip/
        - echo "Hello, world!" > $READTHEDOCS_OUTPUT/htmlzip/index.zip
    post_build:
      - curl -X POST \
        -F "project=${READTHEDOCS_PROJECT}" \
        -F "version=${READTHEDOCS_VERSION}" https://example.com/webhooks/readthedocs/

Features and limitations

The current working directory is at the root of your project’s cloned repository.
Environment variables are expanded for each individual command (see Environment variable reference).
Each command is executed in a new shell process, so modifications done to the shell environment do not persist between commands.
Any command returning non-zero exit code will cause the build to fail immediately (note there is a special exit code to cancel the build).
build.os and build.tools are required when using build.jobs.
If the sphinx or mkdocs configuration is defined, the create_environment, install, and build jobs will use the default commands for the selected tool.
If neither of the sphinx or mkdocs configurations are defined, the create_environment, install, and build jobs will default to run nothing, giving you full control over the build process.

Where to put files

It is your responsibility to generate HTML and other formats of your documentation when overriding the steps from build.jobs.build. The contents of the $READTHEDOCS_OUTPUT/<format>/ directory will be hosted as part of your documentation.

We store the the base folder name _readthedocs/ in the environment variable $READTHEDOCS_OUTPUT and encourage that you use this to generate paths.

Supported formats are published if they exist in the following directories:

$READTHEDOCS_OUTPUT/html/ (required)
$READTHEDOCS_OUTPUT/htmlzip/
$READTHEDOCS_OUTPUT/pdf/
$READTHEDOCS_OUTPUT/epub/

Note

Remember to create the folders before adding content to them. You can ensure that the output folder exists by adding the following command:

mkdir -p $READTHEDOCS_OUTPUT/html/

Search support

Read the Docs will automatically index the content of all your HTML files, respecting the search option.

You can access the search from the Read the Docs dashboard, or by using the Server side search API.

Note

In order for Read the Docs to index your HTML files correctly, they should follow the conventions described at Server side search integration.

Alternative syntax

Alternatively, you can use the build.commands key to completely override the build process.

.readthedocs.yaml

version: 2
build:
  os: "ubuntu-22.04"
  tools:
    python: "3.10"
  commands:
    - pip install pelican
    - pelican --settings docs/pelicanconf.py --output $READTHEDOCS_OUTPUT/html/ docs/

But we recommend using build.jobs instead:

.readthedocs.yaml

version: 2
build:
  os: "ubuntu-22.04"
  tools:
    python: "3.10"
  jobs:
    install:
      - pip install pelican
    build:
      html:
        - pelican --settings docs/pelicanconf.py --output $READTHEDOCS_OUTPUT/html/ docs/

build.jobs offers the same functionality as build.commands, but in a more structured way that allows you to define different commands for each format, while also supporting installing system dependencies via build.apt_packages.

Examples

We’ve included some common examples where using build.jobs will be useful. These examples may require some adaptation for each projects’ use case, we recommend you use them as a starting point.

Unshallow git clone

Read the Docs does not perform a full clone in the checkout job in order to reduce network data and speed up the build process. Instead, it performs a shallow clone and only fetches the branch or tag that you are building documentation for. Because of this, extensions that depend on the full Git history will fail. To avoid this, it’s possible to unshallow the git clone:

.readthedocs.yaml

version: 2
build:
  os: "ubuntu-20.04"
  tools:
    python: "3.10"
  jobs:
    post_checkout:
      - git fetch --unshallow || true

If your build also relies on the contents of other branches, it may also be necessary to re-configure git to fetch these:

.readthedocs.yaml

version: 2
build:
  os: "ubuntu-20.04"
  tools:
    python: "3.10"
  jobs:
    post_checkout:
      - git fetch --unshallow || true
      - git config remote.origin.fetch '+refs/heads/*:refs/remotes/origin/*' || true
      - git fetch --all --tags || true

Cancel build based on a condition

When a command exits with code 183, Read the Docs will cancel the build immediately. You can use this approach to cancel builds that you don’t want to complete based on some conditional logic.

Note

Why 183 was chosen for the exit code?

It’s the word “skip” encoded in ASCII. Then it’s taken the 256 modulo of it because the Unix implementation does this automatically for exit codes greater than 255.

>>> sum(list("skip".encode("ascii")))
439
>>> 439 % 256
183

Here is an example that cancels builds from pull requests when there are no changes to the docs/ folder compared to the origin/main branch:

.readthedocs.yaml

version: 2
build:
  os: "ubuntu-22.04"
  tools:
    python: "3.12"
  jobs:
    post_checkout:
      # Cancel building pull requests when there aren't changed in the docs directory or YAML file.
      # You can add any other files or directories that you'd like here as well,
      # like your docs requirements file, or other files that will change your docs build.
      #
      # If there are no changes (git diff exits with 0) we force the command to return with 183.
      # This is a special exit code on Read the Docs that will cancel the build immediately.
      - |
        if [ "$READTHEDOCS_VERSION_TYPE" = "external" ] && git diff --quiet origin/main -- docs/ .readthedocs.yaml;
        then
          exit 183;
        fi

This other example shows how to cancel a build if the commit message contains skip ci on it:

.readthedocs.yaml

version: 2
build:
  os: "ubuntu-22.04"
  tools:
    python: "3.12"
  jobs:
    post_checkout:
      # Use `git log` to check if the latest commit contains "skip ci",
      # in that case exit the command with 183 to cancel the build
      - (git --no-pager log --pretty="tformat:%s -- %b" -1 | paste -s -d " " | grep -viq "skip ci") || exit 183

Generate documentation from annotated sources with Doxygen

It’s possible to run Doxygen as part of the build process to generate documentation from annotated sources:

.readthedocs.yaml

version: 2
build:
  os: "ubuntu-20.04"
  tools:
    python: "3.10"
  jobs:
    pre_build:
    # Note that this HTML won't be automatically uploaded,
    # unless your documentation build includes it somehow.
      - doxygen

Use MkDocs extensions with extra required steps

There are some MkDocs extensions that require specific commands to be run to generate extra pages before performing the build. For example, pydoc-markdown

.readthedocs.yaml

version: 2
mkdocs:
  configuration: mkdocs.yml
build:
  os: "ubuntu-20.04"
  tools:
    python: "3.10"
  jobs:
    pre_build:
      - pydoc-markdown --build --site-dir "$READTHEDOCS_OUTPUT/html"

Avoid having a dirty Git index

Read the Docs needs to modify some files before performing the build to be able to integrate with some of its features. Because of this reason, it could happen the Git index gets dirty (it will detect modified files). In case this happens and the project is using any kind of extension that generates a version based on Git metadata (like setuptools_scm), this could cause an invalid version number to be generated. In that case, the Git index can be updated to ignore the files that Read the Docs has modified.

.readthedocs.yaml

version: 2
build:
  os: "ubuntu-20.04"
  tools:
    python: "3.10"
  jobs:
    pre_install:
      - git update-index --assume-unchanged environment.yml docs/conf.py

Perform a check for broken links

Sphinx comes with a linkcheck builder that checks for broken external links included in the project’s documentation. This helps ensure that all external links are still valid and readers aren’t linked to non-existent pages.

.readthedocs.yaml

version: 2
build:
  os: "ubuntu-20.04"
  tools:
    python: "3.10"
  jobs:
    pre_build:
      - python -m sphinx -b linkcheck -D linkcheck_timeout=1 docs/ $READTHEDOCS_OUTPUT/linkcheck

Support Git LFS (Large File Storage)

In case the repository contains large files that are tracked with Git LFS, there are some extra steps required to be able to download their content. It’s possible to use post_checkout user-defined job for this.

.readthedocs.yaml

version: 2
build:
  os: "ubuntu-20.04"
  tools:
    python: "3.10"
  jobs:
    post_checkout:
      # Download and uncompress the binary
      # https://git-lfs.github.com/
      - wget https://github.com/git-lfs/git-lfs/releases/download/v3.1.4/git-lfs-linux-amd64-v3.1.4.tar.gz
      - tar xvfz git-lfs-linux-amd64-v3.1.4.tar.gz
      # Modify LFS config paths to point where git-lfs binary was downloaded
      - git config filter.lfs.process "`pwd`/git-lfs filter-process"
      - git config filter.lfs.smudge  "`pwd`/git-lfs smudge -- %f"
      - git config filter.lfs.clean "`pwd`/git-lfs clean -- %f"
      # Make LFS available in current repository
      - ./git-lfs install
      # Download content from remote
      - ./git-lfs fetch
      # Make local files to have the real content on them
      - ./git-lfs checkout

Install Node.js dependencies

It’s possible to install Node.js together with the required dependencies by using user-defined build jobs. To setup it, you need to define the version of Node.js to use and install the dependencies by using build.jobs.post_install:

.readthedocs.yaml

version: 2
build:
  os: "ubuntu-22.04"
  tools:
    python: "3.9"
    nodejs: "16"
  jobs:
    post_install:
      # Install dependencies defined in your ``package.json``
      - npm ci
      # Install any other extra dependencies to build the docs
      - npm install -g jsdoc

Install dependencies with Poetry

Projects managed with Poetry, can use the post_create_environment user-defined job to use Poetry for installing Python dependencies. Take a look at the following example:

.readthedocs.yaml

version: 2

build:
  os: "ubuntu-22.04"
  tools:
    python: "3.10"
  jobs:
    post_install:
      # Install poetry
      # https://python-poetry.org/docs/#installing-manually
      - pip install poetry
      # Install dependencies with 'docs' dependency group
      # https://python-poetry.org/docs/managing-dependencies/#dependency-groups
      # VIRTUAL_ENV needs to be set manually for now.
      # See https://github.com/readthedocs/readthedocs.org/pull/11152/
      - VIRTUAL_ENV=$READTHEDOCS_VIRTUALENV_PATH poetry install --with docs

sphinx:
  configuration: docs/conf.py

Install dependencies with `uv`

Projects can use uv, to install Python dependencies, usually reducing the time taken to install compared to pip. Take a look at the following example:

.readthedocs.yaml

version: 2

build:
   os: ubuntu-24.04
   tools:
      python: "3.13"
   jobs:
      create_environment:
         - asdf plugin add uv
         - asdf install uv latest
         - asdf global uv latest
         - uv venv
      install:
         - uv pip install -r requirements.txt
      build:
         html:
            - uv run sphinx-build -T -b html docs $READTHEDOCS_OUTPUT/html

MkDocs projects could use NO_COLOR=1 uv run mkdocs build --strict --site-dir $READTHEDOCS_OUTPUT/html instead.

Update Conda version

Projects using Conda may need to install the latest available version of Conda. This can be done by using the pre_create_environment user-defined job to update Conda before creating the environment. Take a look at the following example:

.readthedocs.yaml

 version: 2

 build:
   os: "ubuntu-22.04"
   tools:
     python: "miniconda3-4.7"
   jobs:
     pre_create_environment:
       - conda update --yes --quiet --name=base --channel=defaults conda

sphinx:
   configuration: docs/conf.py

 conda:
   environment: environment.yml

Using Pelican

Pelican is a well-known static site generator that’s commonly used for blogs and landing pages. If you are building your project with Pelican you could use a configuration file similar to the following:

.readthedocs.yaml

version: 2
build:
  os: "ubuntu-22.04"
  tools:
    python: "3.10"
  jobs:
    install:
      - pip install pelican[markdown]
    build:
      html:
        - pelican --settings docs/pelicanconf.py --output $READTHEDOCS_OUTPUT/html/ docs/

Using Docsify

Docsify generates documentation websites on the fly, without the need to build static HTML. These projects can be built using a configuration file like this:

.readthedocs.yaml

version: 2
build:
  os: "ubuntu-22.04"
  jobs:
    build:
      html:
        - mkdir --parents $READTHEDOCS_OUTPUT/html/
        - cp --recursive docs/* $READTHEDOCS_OUTPUT/html/

Using Asciidoc

Asciidoctor is a fast processor for converting and generating documentation from AsciiDoc source. The Asciidoctor toolchain includes Asciidoctor.js which you can use with custom build commands. Here is an example configuration file:

.readthedocs.yaml

version: 2
build:
  os: "ubuntu-22.04"
  tools:
    nodejs: "20"
  jobs:
    install:
      - npm install -g asciidoctor
    build:
      html:
        - asciidoctor -D $READTHEDOCS_OUTPUT/html index.asciidoc

Build process customization

Extend or override the build process

Features and limitations

Where to put files

Search support

Alternative syntax

Examples

Unshallow git clone

Cancel build based on a condition

Generate documentation from annotated sources with Doxygen

Use MkDocs extensions with extra required steps

Avoid having a dirty Git index

Perform a check for broken links

Support Git LFS (Large File Storage)

Install Node.js dependencies

Install dependencies with Poetry

Install dependencies with uv

Update Conda version

Using Pelican

Using Docsify

Using Asciidoc

Install dependencies with `uv`