Config-based Git hooks

Git hooks let you run scripts automatically on commit, push, or merge. But they have always had a few rough edges: the scripts live in .git/hooks/, which is never committed to the repository, each event can only have one script, and there is no built-in way to see what hooks are active or where they come from.

Git 2.54 addresses all three with config-based hooks. You can now define hooks directly in your Git configuration files.

# Config-based hooks in Git 2.54

Instead of placing an executable script at .git/hooks/pre-commit, you can now write this in your .gitconfig or .git/config:

[hook "lint"]
    event = pre-commit
    command = ./hooks/pre-commit-lint

The hook.<name>.event key specifies which Git event triggers the hook, and hook.<name>.command is the command to run. The name (lint here) is just a label, it can be anything you like.

Because this is standard Git config, it can live in any of the usual config scopes:

• - ~/.gitconfig, applies to all your repos globally
• - .git/config, applies to a single repository only
• - /etc/gitconfig, applies system-wide

# Multiple hooks per event

Previously, each hook event could only have one script. If you wanted to run both a linter and a secret scanner before a commit, you needed a wrapper script that called both. With config-based hooks, you just define them separately:

[hook "lint"]
    event = pre-commit
    command = ./hooks/pre-commit-lint

[hook "no-leaks"]
    event = pre-commit
    command = ./hooks/pre-commit-secrets

Git runs them in the order it encounters them in the config. The traditional script at .git/hooks/pre-commit still works too and runs last, so existing setups are unaffected.

$ git hook list pre-commit
global    lint      ./hooks/pre-commit-lint
local     no-leaks  ./hooks/pre-commit-secrets

# Setting it up for a team

The actual hook scripts should be committed to the repository in a hooks/ directory. Make sure each script has the executable bit set:

mkdir hooks
chmod +x hooks/pre-commit-lint

Then each team member needs to register the hooks in their local .git/config after cloning. This is two commands per hook:

git config --local hook.lint.event pre-commit
git config --local hook.lint.command ./hooks/pre-commit-lint

Put these commands in your project's setup script — a Makefile, justfile, or a simple setup.sh — so new team members only need to run it once after cloning.

# Makefile
setup:
	git config --local hook.lint.event pre-commit
	git config --local hook.lint.command ./hooks/pre-commit-lint

# Disabling a hook per repository

If a hook is defined globally in ~/.gitconfig but you want to opt a specific repository out, you can disable it locally without removing the global config:

git config --local hook.lint.enabled false

This is particularly useful for personal global hooks like a secrets scanner or a commit message linter that you want everywhere except in a specific repo.

Config-based hooks are one of those small quality-of-life improvements that quietly make Git more self-contained. The hooks themselves are now first-class config: composable, inspectable with git hook list, and easy to disable per repo without touching anything globally.