About

git-cliff can generate changelog files from the Git history by utilizing conventional commits as well as regex-powered custom parsers. The changelog template can be customized with a configuration file to match the desired format.

Table of Contents

• Installation
  • From crates.io
  • Using pacman
  • From NPM
  • From MacPorts
  • Binary releases
  • Build from source
• Usage
  • Command Line Arguments
  • Examples
• Docker
• GitHub Actions
  • git-cliff-action
  • setup-git-cliff
• GitLab CI/CD
• Configuration File
  • changelog
    • header
    • body
    • trim
    • footer
  • git
    • conventional_commits
    • filter_unconventional
    • split_commits
    • commit_preprocessors
    • commit_parsers
    • protect_breaking_commits
    • filter_commits
    • tag_pattern
    • skip_tags
    • ignore_tags
    • topo_order
    • sort_commits
    • link_parsers
    • limit_commits
• Project Integration
  • Rust
  • Python
• Templating
  • Context
    • Conventional Commits
      • Footers
      • Breaking Changes
      • Committer vs Author
    • Non-Conventional Commits
  • Syntax
  • Examples
• In The Media
• Similar/Related Projects
• License
• Copyright

Installation

Packaging status

From crates.io

git-cliff can be installed from crates.io:

cargo install git-cliff

Minimum supported Rust version is 1.60.0.

Using pacman

If you are using Arch Linux, git-cliff can be installed from the community repository:

pacman -S git-cliff

From NPM

You can install and run git-cliff with a single command:

npx git-cliff@latest

Also, if you want to add git-cliff to your project:

# with yarn
yarn add -D git-cliff

# with npm
npm install git-cliff --save-dev

Afterwards, you can run git-cliff via npm exec git-cliff or npx git-cliff@latest.

From MacPorts

On macOS, git-cliff can be installed via MacPorts:

sudo port install git-cliff

Port page

Binary releases

See the available binaries for different operating systems/architectures from the releases page.

* Release tarballs are signed with the following PGP key: 1D2D410A741137EBC544826F4A92FA17B6619297

Build from source

• Linux dependencies: zlib

# binary will be located at `target/release/git-cliff`
CARGO_TARGET_DIR=target cargo build --release

Usage

Command Line Arguments

git-cliff [FLAGS] [OPTIONS] [--] [RANGE]

Flags:

-v, --verbose       Increases the logging verbosity
-i, --init          Writes the default configuration file to cliff.toml
-l, --latest        Processes the commits starting from the latest tag
    --current       Processes the commits that belong to the current tag
-u, --unreleased    Processes the commits that do not belong to a tag
    --topo-order    Sorts the tags topologically
    --context       Prints changelog context as JSON
-h, --help          Prints help information
-V, --version       Prints version information

Options:

-c, --config <PATH>                Sets the configuration file [env: GIT_CLIFF_CONFIG=] [default: cliff.toml]
-w, --workdir <PATH>               Sets the working directory [env: GIT_CLIFF_WORKDIR=]
-r, --repository <PATH>...         Sets the git repository [env: GIT_CLIFF_REPOSITORY=]
    --include-path <PATTERN>...    Sets the path to include related commits [env: GIT_CLIFF_INCLUDE_PATH=]
    --exclude-path <PATTERN>...    Sets the path to exclude related commits [env: GIT_CLIFF_EXCLUDE_PATH=]
    --with-commit <MSG>...         Sets custom commit messages to include in the changelog [env: GIT_CLIFF_WITH_COMMIT=]
-p, --prepend <PATH>               Prepends entries to the given changelog file [env: GIT_CLIFF_PREPEND=]
-o, --output <PATH>                Writes output to the given file [env: GIT_CLIFF_OUTPUT=]
-t, --tag <TAG>                    Sets the tag for the latest version [env: GIT_CLIFF_TAG=]
-b, --body <TEMPLATE>              Sets the template for the changelog body [env: GIT_CLIFF_TEMPLATE=]
-s, --strip <PART>                 Strips the given parts from the changelog [possible values: header, footer, all]
    --sort <SORT>                  Sets sorting of the commits inside sections [default: oldest] [possible values: oldest, newest]

Args:

<RANGE>    Sets the commit range to process

Examples

The default configuration file (cliff.toml) can be generated using the --init flag:

# create cliff.toml
git cliff --init

Then simply create a changelog at your projects git root directory:

# same as running `git-cliff --config cliff.toml --repository .`
# same as running `git-cliff --workdir .`
git cliff

Set a tag for the "unreleased" changes:

# it doesn't have to be an existing tag
git cliff --tag 1.0.0

Generate a changelog for a certain part of git history:

# only takes the latest tag into account
git cliff --latest

# only takes the current tag into account
# useful if you checkout a specific tag (e.g. `git checkout v0.0.1`)
# (requires a tag to be present for the current commit (i.e. HEAD))
git cliff --current

# generate changelog for unreleased commits
git cliff --unreleased
git cliff --unreleased --tag 1.0.0

# generate changelog for a specific commit range
git cliff 4c7b043..a440c6e
git cliff 4c7b043..HEAD
git cliff HEAD~2..

Generate a changelog scoped to a specific directory (useful for monorepos):

git cliff --include-path "**/*.toml" --include-path "*.md"
git cliff --exclude-path ".github/*"

Generate a changelog for multiple git repositories:

# merges the commit history
git cliff --repository path1 path2

Generate a changelog that includes yet unexisting commit messages:

commit_msg="chore(release): update CHANGELOG.md for 1.0.0"

# You might need to include the commit messages that do not exist
# for testing purposes or solving the chicken-egg problem.
git cliff --with-commit "$commit_msg" -o CHANGELOG.md

git add CHANGELOG.md && git commit -m "$commit_msg"

The commit SHA will be empty as default when --with-commit is used. Specify the hash with a message separated by single whitespace for setting the commit SHA. e.g. --with-commit "8f55e69eba6e6ce811ace32bd84cc82215673cb6 feat: add X"

Sort the commits inside sections:

# The oldest commit will be on top.
# (default)
git cliff --sort oldest

# The newest commit will be on top.
git cliff --sort newest

Sort the tags in topological order:

# Process in topological order instead of chronological.
git cliff --topo-order

Save the changelog file to the specified file:

git cliff --output CHANGELOG.md

Print the changelog context as JSON:

# print context to stdout
git cliff --context

# save context to a file
git cliff --context --output context.json

Prepend new changes to an existing changelog file:

# 1- changelog header is removed from CHANGELOG.md
# 2- new entries are prepended to CHANGELOG.md without footer part
git cliff --unreleased --tag 1.0.0 --prepend CHANGELOG.md

Set/remove the changelog parts:

git cliff --body $template --strip footer

Also, see the release script of this project which sets the changelog as a message of an annotated tag.

Docker

The easiest way of running git-cliff (in the git root directory with configuration file present) is to use the available tags from Docker Hub:

docker run -t -v "$(pwd)/.git":/app/ orhunp/git-cliff:latest

Or you can use the image from the GitHub Package Registry:

docker run -t -v "$(pwd)/.git":/app/ docker.pkg.github.com/orhun/git-cliff/git-cliff:latest

Also, you can build the image yourself using DOCKER_BUILDKIT=1 docker build -t git-cliff . command.

GitHub Actions

git-cliff-action

It is possible to generate changelogs using GitHub Actions via git-cliff-action.

- name: Generate a changelog
  uses: orhun/git-cliff-action@v2
  with:
    config: cliff.toml
    args: --verbose
  env:
    OUTPUT: CHANGELOG.md

See the repository for other examples.

Also, see the continuous deployment workflow of this project which sets the release notes for GitHub releases using this action.

setup-git-cliff

There is also another GitHub Action which is setup-git-cliff.

While git-cliff-action uses the Docker image generated by docker.yml, setup-git-cliff installs the binary executable in the release artifacts:

- name: Check out repository
  uses: actions/checkout@v3

- name: Set up git-cliff
  uses: kenji-miyake/setup-git-cliff@v1

- name: Run git-cliff
  run: |
    git cliff    

See a practical example here.

GitLab CI/CD

It is possible to generate changelogs using GitLab CI/CD.

This minimal example creates artifacts that can be used on another job.

- changelog:
    image:
      name: orhunp/git-cliff:latest
      entrypoint: [""]
    variables:
      GIT_STRATEGY: clone # clone entire repo instead of reusing workspace
      GIT_DEPTH: 0 # avoid shallow clone to give cliff all the info it needs
    stage: doc
    script:
      - git-cliff -r . > CHANGELOG.md
    artifacts:
      paths:
        - CHANGELOG.md

Please note that the stage is doc and has to be changed accordingly to your need.

Configuration File

git-cliff configuration file supports TOML (preferred) and YAML formats.

The configuration file is read from $HOME/git-cliff/cliff.toml if the file exists. This location depends on the platform, for example:

• on Linux: /home/<user>/.config/git-cliff/cliff.toml
• on Windows: C:\Users\<user>\AppData\Roaming\git-cliff\cliff.toml
• on macOS: /Users/<user>/Library/Application Support/git-cliff/cliff.toml

See config/cliff.toml for the default configuration values.

changelog

This section contains the configuration options for changelog generation.

[changelog]
header = "Changelog"
body = """
{% for group, commits in commits | group_by(attribute="group") %}
    ### {{ group | upper_first }}
    {% for commit in commits %}
        - {{ commit.message | upper_first }}
    {% endfor %}
{% endfor %}
"""
trim = true
footer = "<!-- generated by git-cliff -->"

header

Header text that will be added to the beginning of the changelog.

body

Body template that represents a single release in the changelog.

See templating for more detail.

trim

If set to true, leading and trailing whitespace are removed from the body.

It is useful for adding indentation to the template for readability, as shown in the example.

footer

Footer text that will be added to the end of the changelog.

git

This section contains the parsing and git related configuration options.

[git]
conventional_commits = true
filter_unconventional = true
split_commits = false
commit_parsers = [
    { message = "^feat", group = "Features"},
    { message = "^fix", group = "Bug Fixes"},
    { message = "^doc", group = "Documentation"},
    { message = "^perf", group = "Performance"},
    { message = "^refactor", group = "Refactor"},
    { message = "^style", group = "Styling"},
    { message = "^test", group = "Testing"},
]
protect_breaking_commits = false
filter_commits = false
tag_pattern = "v[0-9]*"
skip_tags = "v0.1.0-beta.1"
ignore_tags = ""
topo_order = false
sort_commits = "oldest"
link_parsers = [
    { pattern = "#(\\d+)", href = "https://github.com/orhun/git-cliff/issues/$1"},
    { pattern = "RFC(\\d+)", text = "ietf-rfc$1", href = "https://datatracker.ietf.org/doc/html/rfc$1"},
]
limit_commits = 42

conventional_commits

If set to true, commits are parsed according to the Conventional Commits specifications.

The Conventional Commits specification is a lightweight convention on top of commit messages. It provides an easy set of rules for creating an explicit commit history; which makes it easier to write automated tools on top of. This convention dovetails with SemVer, by describing the features, fixes, and breaking changes made in commit messages.

The commit message should be structured as follows:

<type>[optional scope]: <description>

[optional body]

[optional footer(s)]

e.g. feat(parser): add ability to parse arrays

filter_unconventional

If set to true, commits that are not conventional are excluded. This option can be used to generate changelogs with conventional and unconventional commits mixed together. For example:

conventional_commits = true
filter_unconventional = false
commit_parsers = [
  { message = ".*", group = "Other", default_scope = "other"},
]

With the configuration above, conventional commits are parsed as usual and unconventional commits will be also included in the changelog as "Other".

To completely exclude unconventional commits from the changelog:

# default behaviour
conventional_commits = true
filter_unconventional = true

To include any type of commit in the changelog without parsing:

conventional_commits = false
filter_unconventional = false

split_commits

This flag violates "conventional commits". It should remain off by default if conventional commits is to be respected.

If set to true, each line of a commit is processed individually, as if it were its own commit message. This may cause a commit to appear multiple times in a changelog, once for each match.

conventional_commits = true
filter_unconventional = true
split_commits = true
commit_parsers = [
    { message = "^feat", group = "Features"},
]

With the configuration above, lines are parsed as conventional commits and unconventional lines are omitted.

If filter_unconventional = false, every line will be processes as an unconventional commit, resulting in each line of a commit being treated as a changelog entry.

commit_preprocessors

An array of commit preprocessors for manipulating the commit messages before parsing/grouping them. These regex-based preprocessors can be used for removing or selecting certain parts of the commit message/body to be used in the following processes.

Examples:

• { pattern = "foo", replace = "bar"}
  • Replace text.
• { pattern = 'Merged PR #[0-9]: (.*)', replace = "$1"}
  • Remove prefix.
• { pattern = " +", replace = " "}
  • Replace multiple spaces with a single space.
• { pattern = "\\(#([0-9]+)\\)", replace = "([#${1}](https://github.com/orhun/git-cliff/issues/${1}))"}
  • Replace the issue number with the link.
• { pattern = "https://github.com/[^ ]/issues/([0-9]+)", replace = "[Issue #${1}]"}
  • Replace the issue link with the number.
• { pattern = "Merge pull request #([0-9]+) from [^ ]+", replace = "PR # [${1}](https://github.com/orhun/git-cliff/pull/${1}):"}
  • Hyperlink PR references from merge commits.
• { pattern = "https://github.com/orhun/git-cliff/commit/([a-f0-9]{7})[a-f0-9]*", replace = "commit # [${1}](${0})"}
  • Hyperlink commit links, with short commit hash as description.
• { pattern = "([ \\n])(([a-f0-9]{7})[a-f0-9]*)", replace = "${1}commit # [${3}](https://github.com/orhun/git-cliff/commit/${2})"}
  • Hyperlink bare commit hashes like "abcd1234" in commit logs, with short commit hash as description.

Custom OS commands can also be used for modifying the commit messages:

• { pattern = "foo", replace_command = "pandoc -t commonmark"}

This is useful when you want to filter/encode messages using external commands. In the example above, pandoc is used to convert each commit message that matches the given pattern to the CommonMark format.

A more fun example would be reversing each commit message:

• { pattern = '.*', replace_command = 'rev | xargs echo "reversed: $@"' }

$COMMIT_SHA environment variable is set during execution of the command so you can do fancier things like reading the commit itself:

• { pattern = '.*', replace_command = 'git show -s --format=%B $COMMIT_SHA' }

commit_parsers

An array of commit parsers for determining the commit groups by using regex.

Examples:

• { message = "^feat", group = "Features"}
  • Group the commit as "Features" if the commit message (description) starts with "feat".
• { body = ".*security", group = "Security"}
  • Group the commit as "Security" if the commit body contains "security".
• { message = '^fix\((.*)\)', group = 'Fix (${1})' }
  • Use the matched scope value from the commit message in the group name.
• { message = ".*deprecated", body = ".*deprecated", group = "Deprecation"}
  • Group the commit as "Deprecation" if the commit body and message contains "deprecated".
• { message = "^revert", skip = true}
  • Skip processing the commit if the commit message (description) starts with "revert".
• { message = "^doc", group = "Documentation", default_scope = "other"},
  • If the commit starts with "doc", group the commit as "Documentation" and set the default scope to "other". (e.g. docs: xyz will be processed as docs(other): xyz)
• { message = "(www)", scope = "Application"}
  • If the commit contains "(www)", override the scope with "Application". Scoping order is: scope specification, conventional commit's scope and default scope.

protect_breaking_commits

If set to true, any breaking changes will be protected against being skipped due to any commit parser.

filter_commits

If set to true, commits that are not matched by commit parsers are filtered out.

tag_pattern

A glob pattern for matching the git tags.

e.g. It processes the same tags as the output of the following git command:

git tag --list 'v[0-9]*'

skip_tags

A regex for skip processing the matched tags.

ignore_tags

A regex for ignore processing the matched tags.

While skip_tags drop commits from the changelog, ignore_tags include ignored commits into the next tag.

topo_order

If set to true, tags are processed in topological order instead of chronological.

This can also be achieved by using the --topo-order command line flag.

sort_commits

Sort the commits inside sections by specified order.

Possible values:

• oldest
• newest

This can also be achieved by specifying the --sort command line argument.

link_parsers

An array of link parsers for extracting external references, and turning them into URLs, using regex.

Examples:

• { pattern = "#(\\d+)", href = "https://github.com/orhun/git-cliff/issues/$1"}
  • Extract all GitLab issues and PRs and generate URLs linking to them. The link text will be the matching pattern.
• { pattern = "RFC(\\d+)", text = "ietf-rfc$1", href = "https://datatracker.ietf.org/doc/html/rfc$1"},
  • Extract mentions of IETF RFCs and generate URLs linking to them. It also rewrites the text as "ietf-rfc...".

limit_commits

limit_commits is a optional positive integer number that limits the number of included commits in the generated changelog.

limit_commits is not part of the default configuration.

Project Integration

Rust

For Rust projects, git-cliff can be configured in Cargo.toml via metadata table. To do this, simply replace the available configuration sections with [package.metadata.git-cliff.<section>] and place them inside Cargo.toml. For example:

[package]
name = "..."

[dependencies]
# ...

[package.metadata.git-cliff.changelog]
header = "All notable changes to this project will be documented in this file."
body = "..."
footer = "<!-- generated by git-cliff -->"
# see [changelog] section for more keys

[package.metadata.git-cliff.git]
conventional_commits = true
commit_parsers = []
filter_commits = false
# see [git] section for more keys

For Cargo workspaces, workspace.metadata table can be used. (e.g. [workspace.metadata.git-cliff.<section>])

Python

For Python projects, git-cliff can be configured in pyproject.toml via the tool table. To do this, simply replace the available configuration sections with [tool.git-cliff.<section>] and place them inside pyproject.toml. For example:

name = "..."

[project]
dependencies = []

[tool.git-cliff.changelog]
header = "All notable changes to this project will be documented in this file."
body = "..."
footer = "<!-- generated by git-cliff -->"
# see [changelog] section for more keys
[tool.git-cliff.git]
conventional_commits = true
commit_parsers = []
filter_commits = false
# see [git] section for more keys

Templating

A template is a text where variables and expressions get replaced with values when it is rendered.

Context

Context is the model that holds the required data for a template rendering. The JSON format is used in the following examples for the representation of a context.

Conventional Commits

conventional_commits = true

For a conventional commit like this,

<type>[scope]: <description>

[body]

[footer(s)]

following context is generated to use for templating:

{
  "version": "v0.1.0-rc.21",
  "commits": [
    {
      "id": "e795460c9bb7275294d1fa53a9d73258fb51eb10",
      "group": "<type> (overrided by commit_parsers)",
      "scope": "[scope]",
      "message": "<description>",
      "body": "[body]",
      "footers": [
        {
          "token": "<name of the footer, such as 'Signed-off-by'>",
          "separator": "<the separator between the token and value, such as ':'>",
          "value": "<the value following the separator",
          "breaking": false
        }
      ],
      "breaking_description": "<description>",
      "breaking": false,
      "conventional": true,
      "links": [
        { "text": "(set by link_parsers)", "href": "(set by link_parsers)" }
      ],
      "author": {
        "name": "User Name",
        "email": "user.email@example.com",
        "timestamp": 1660330071
      },
      "committer": {
        "name": "User Name",
        "email": "user.email@example.com",
        "timestamp": 1660330071
      }
    }
  ],
  "commit_id": "a440c6eb26404be4877b7e3ad592bfaa5d4eb210 (release commit)",
  "timestamp": 1625169301,
  "previous": {
    "version": "previous release"
  }
}

Footers

A conventional commit's body may end with any number of structured key-value pairs known as footers. These consist of a string token naming the footer, a separator (which is either : or #), and a value, similar to the git trailers convention.

For example:

• Signed-off-by: User Name <user.email@example.com>
• Reviewed-by: User Name <user.email@example.com>
• Fixes #1234
• BREAKING CHANGE: breaking change description

When a conventional commit contains footers, the footers are passed to the template in a footers array in the commit object. Each footer is represented by an object with the following fields:

• token, the name of the footer (preceding the separator character)
• separator, the footer's separator string (either : or #)
• value, the value following the separator character
• breaking, which is true if this is a BREAKING CHANGE: footer, and false otherwise

Breaking Changes

breaking flag is set to true when the commit has an exclamation mark after the commit type and scope, e.g.:

feat(scope)!: this is a breaking change

Or when the BREAKING CHANGE: footer is defined:

feat: add xyz

BREAKING CHANGE: this is a breaking change

breaking_description is set to the explanation of the breaking change. This description is expected to be present in the BREAKING CHANGE footer. However, if it's not provided, the message is expected to describe the breaking change.

If the BREAKING CHANGE: footer is present, the footer will also be included in commit.footers.

Breaking changes will be skipped if protect_breaking_commits is set to true, even when matched by a skipping commit_parser.

Committer vs Author

From Git docs:

You may be wondering what the difference is between author and committer. The author is the person who originally wrote the work, whereas the committer is the person who last applied the work. So, if you send in a patch to a project and one of the core members applies the patch, both of you get credit — you as the author, and the core member as the committer.

Non-Conventional Commits

conventional_commits = false

If conventional_commits is set to false, then some of the fields are omitted from the context or squashed into the message field:

{
  "version": "v0.1.0-rc.21",
  "commits": [
    {
      "id": "e795460c9bb7275294d1fa53a9d73258fb51eb10",
      "group": "(overrided by commit_parsers)",
      "scope": "(overrided by commit_parsers)",
      "message": "(full commit message including description, footers, etc.)",
      "conventional": false,
      "links": [
        { "text": "(set by link_parsers)", "href": "(set by link_parsers)" }
      ],
      "author": {
        "name": "User Name",
        "email": "user.email@example.com",
        "timestamp": 1660330071
      },
      "committer": {
        "name": "User Name",
        "email": "user.email@example.com",
        "timestamp": 1660330071
      }
    }
  ],
  "commit_id": "a440c6eb26404be4877b7e3ad592bfaa5d4eb210 (release commit)",
  "timestamp": 1625169301,
  "previous": {
    "version": "previous release"
  }
}

Syntax

git-cliff uses Tera as the template engine. It has a syntax based on Jinja2 and Django templates.

There are 3 kinds of delimiters and those cannot be changed:

• {{ and }} for expressions
• {% or {%- and %} or -%} for statements
• {# and #} for comments

See the Tera Documentation for more information about control structures, built-ins filters, etc.

Custom built-in filters that git-cliff uses:

• upper_first: Converts the first character of a string to uppercase.

Examples

See EXAMPLES.md.

In The Media

• git-cliff was featured in Console #141 - The Open Source Newsletter

Similar/Related Projects

• git-journal - The Git Commit Message and Changelog Generation Framework
• clog-cli - Generate beautiful changelogs from your Git commit history
• relnotes - A tool to automatically generate release notes for your project.
• cocogitto - A set of CLI tools for the conventional commit and semver specifications.
• cliff-jumper - A NodeJS CLI tool that combines git-cliff and conventional-recommended-bump to semantically bump a NodeJS package and generate a git-cliff powered changelog.
• release-plz - Release Rust packages from CI.

License

GNU General Public License (v3.0)

Copyright

Copyright © 2021-2023, git-cliff contributors