setup-semantic-release（设置语义发布）

Setup Semantic Release & Conventional Commits

Overview

Set up a fully automated versioning and release pipeline using conventional commits, commitlint, husky git hooks, and semantic-release. Version bumps, changelogs, and GitHub releases are derived automatically from commit messages.

Core principle: Commits drive releases — enforce commit format at author time, automate everything else.

When to Use

Always use when:

• Setting up a new project that needs automated versioning
• Adding conventional commits to an existing repo
• Asked to "set up semantic release" or "add commitlint"
• Migrating from manual versioning to automated releases

Useful for:

• Any npm/Node.js project publishing to npm or GitHub
• Projects that want auto-generated changelogs
• Teams that need consistent commit message formatting

Avoid when:

• Project already has semantic-release configured (check for .releaserc*)
• Project uses a different release tool (changesets, release-it, standard-version)
• Non-Node.js project without package.json (adapt manually instead)

Prerequisites

Before starting, verify:

• Project has a package.json
• Project uses git with a remote on GitHub
• Node.js >= 18 installed
• npm or equivalent package manager available
• A CI/CD environment (GitHub Actions recommended) for automated releases

Setup Workflow

Phase 1: Install Dependencies

Install all required dev dependencies:

npm install --save-dev \
  @commitlint/cli@^19.0.0 \
  @commitlint/config-conventional@^19.0.0 \
  semantic-release@^24.0.0 \
  @semantic-release/changelog@^6.0.0 \
  @semantic-release/git@^10.0.0 \
  husky@^9.0.0

What each package does:

Verification:

• All packages appear in devDependencies
• No install errors

Phase 2: Configure Commitlint

Create commitlint.config.js in the project root:

export default {
  extends: ['@commitlint/config-conventional'],
  rules: {
    'type-enum': [
      2,
      'always',
      [
        'feat',     // New feature
        'fix',      // Bug fix
        'docs',     // Documentation only
        'style',    // Formatting, no code change
        'refactor', // Code change that neither fixes nor adds
        'perf',     // Performance improvement
        'test',     // Adding/updating tests
        'build',    // Build system or dependencies
        'ci',       // CI configuration
        'chore',    // Maintenance tasks
        'revert',   // Revert previous commit
      ],
    ],
    'subject-case': [2, 'always', 'lower-case'],
    'header-max-length': [2, 'always', 100],
    'body-max-line-length': [0], // Disable for semantic-release changelog
  },
};

IMPORTANT: If the project does NOT have "type": "module" in package.json, use module.exports = {...} instead of export default.

Commit message format:

type(scope): subject

body (optional)

footer (optional)

How types map to version bumps:

• feat = minor bump (1.0.0 -> 1.1.0)
• fix = patch bump (1.0.0 -> 1.0.1)
• feat! or BREAKING CHANGE in footer = major bump (1.0.0 -> 2.0.0)
• All other types (docs, test, chore, etc.) = no release

Verification:

• commitlint.config.js exists at project root
• Module syntax matches project type (ESM vs CJS)

Phase 3: Configure Semantic Release

Create .releaserc.json in the project root:

{
  "branches": ["main"],
  "plugins": [
    "@semantic-release/commit-analyzer",
    "@semantic-release/release-notes-generator",
    [
      "@semantic-release/changelog",
      {
        "changelogFile": "CHANGELOG.md"
      }
    ],
    [
      "@semantic-release/git",
      {
        "assets": ["CHANGELOG.md", "package.json", "package-lock.json"],
        "message": "chore(release): ${nextRelease.version} [skip ci]\n\n${nextRelease.notes}"
      }
    ],
    "@semantic-release/github"
  ]
}

Key details:

• branches: Set to your release branch. Change "main" if your default branch is "master" or something else.
• [skip ci] in the release commit message prevents CI from triggering an infinite loop on the release commit.
• The assets array lists files that semantic-release commits back to the repo after a release.
• Plugin order matters — they execute sequentially.

For multi-branch releases (e.g., pre-releases), adjust branches:

{
  "branches": [
    "main",
    { "name": "beta", "prerelease": true },
    { "name": "alpha", "prerelease": true }
  ]
}

Verification:

• .releaserc.json exists at project root
• branches matches the actual default branch name
• Plugin order is correct (analyzer first, github last)

Phase 4: Initialize Husky & Git Hooks

Run husky's init command and add the prepare script:

npx husky init

This creates the .husky/ directory and adds "prepare": "husky" to package.json scripts.

4a. Add the commit-msg hook (required)

Write the commitlint hook:

echo 'npx --no -- commitlint --edit $1' > .husky/commit-msg

This validates every commit message against the commitlint rules before allowing the commit.

4b. Add a pre-commit hook (optional, configurable)

Ask the user what pre-commit hook they want. Common options:

Write the chosen command to the pre-commit hook:

echo '<chosen-command>' > .husky/pre-commit

Or remove the default pre-commit hook if not needed:

rm .husky/pre-commit

Verification:

• .husky/commit-msg exists and contains commitlint command
• .husky/pre-commit configured or removed per user choice
• "prepare": "husky" exists in package.json scripts

Phase 5: Add prepare script (if missing)

Check if package.json already has a prepare script. If npx husky init didn't add it, add manually:

{
  "scripts": {
    "prepare": "husky"
  }
}

This ensures husky hooks are installed automatically when anyone runs npm install.

Phase 6: Create Initial CHANGELOG

Create a starter CHANGELOG.md:

# Changelog

All notable changes to this project will be documented in this file.

The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).

Semantic-release will prepend entries to this file on each release.

Phase 7: CI/CD Setup (GitHub Actions)

Create .github/workflows/release.yml:

name: Release

on:
  push:
    branches: [main]

permissions:
  contents: write
  issues: write
  pull-requests: write

jobs:
  release:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
        with:
          fetch-depth: 0
      - uses: actions/setup-node@v4
        with:
          node-version: 'lts/*'
      - run: npm ci
      - run: npx semantic-release
        env:
          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
          NPM_TOKEN: ${{ secrets.NPM_TOKEN }}

Notes:

• fetch-depth: 0 is required — semantic-release needs full git history to analyze commits.
• GITHUB_TOKEN is provided automatically by GitHub Actions.
• NPM_TOKEN is only needed if publishing to npm. Set it in repo Settings > Secrets. Remove the line if not publishing to npm.
• Update branches if your default branch is not main.

Verification:

• .github/workflows/release.yml exists
• Branch name matches actual default branch
• NPM_TOKEN secret set (if publishing to npm)

Phase 8: Verify Everything Works

Run these checks in order:

# 1. Verify husky hooks are installed
ls .husky/commit-msg

# 2. Test commitlint with a valid message
echo "feat: test message" | npx commitlint

# 3. Test commitlint with an invalid message (should fail)
echo "bad message" | npx commitlint

# 4. Test a real commit (should pass)
git add .
git commit -m "chore: set up semantic release and conventional commits"

# 5. Dry-run semantic-release (optional, requires CI token locally)
npx semantic-release --dry-run

Expected results:

• Step 2: exits 0, no errors
• Step 3: exits non-zero, shows validation errors
• Step 4: commit succeeds
• Step 5: shows what version would be released (if token available)

Quick Reference

Commit Message Cheat Sheet

feat(auth): add login endpoint          # minor bump
fix(api): handle null response           # patch bump
feat!: redesign user model               # MAJOR bump
docs: update readme                      # no release
test: add unit tests for parser          # no release
chore: update dependencies               # no release
refactor(core): simplify error handling  # no release

# With body and breaking change footer:
feat(api): add pagination support

Adds offset/limit parameters to all list endpoints.

BREAKING CHANGE: removed `page` parameter in favor of `offset`

Files Created/Modified

Troubleshooting

Problem: Commitlint rejects valid-looking messages

Cause: Subject starts with uppercase or header exceeds 100 characters.

Solution: Ensure subject is lowercase and under 100 chars total. The subject-case rule enforces lowercase. Example: feat: Add feature fails, feat: add feature passes.

Problem: Husky hooks don't run

Cause: Hooks not installed or .husky/ directory missing.

Solution: Run npx husky init again, then re-create the hook files. Verify "prepare": "husky" is in package.json scripts. Run npm install to trigger the prepare script.

Problem: Semantic-release creates duplicate changelog entries

Cause: The body-max-line-length commitlint rule conflicts with semantic-release's generated notes.

Solution: Set 'body-max-line-length': [0] in commitlint config to disable body line length validation. This is already included in the config above.

Problem: CI release creates infinite loop

Cause: Release commit triggers another CI run which tries to release again.

Solution: The [skip ci] tag in the .releaserc.json git commit message prevents this. Verify it's present in the message field of @semantic-release/git config.

Problem: "ENOGITHEAD" or "EGITNOBRANCH" in CI

Cause: Shallow clone in CI doesn't have full git history.

Solution: Ensure fetch-depth: 0 in the checkout step of GitHub Actions. This fetches full history needed for commit analysis.

Examples

feat(parser): add yaml frontmatter extraction
feat(cli): add interactive selection prompts
fix(writer): handle unicode in file paths
test: add e2e tests for compile workflow
chore(release): 1.0.0 [skip ci]

Each commit has a clear type, optional scope, lowercase subject. Semantic-release can analyze this and produce a clean changelog grouped by type.

Added parser stuff
WIP
fix things
YAML support
updated tests and also fixed a bug and refactored

No types, no scopes, mixed concerns in single commits, uppercase, vague subjects. Commitlint would reject all of these. Semantic-release cannot derive meaningful changelogs.

Integration

setup-semantic-release + generate-skill

When creating new skills for projects that use semantic-release:

• The generated skill's commit conventions should align with the project's commitlint config
• Use feat(skill-name): for the initial skill commit so semantic-release triggers a minor bump

setup-semantic-release + track-session

When setting up semantic-release as a multi-step task:

• Use track-session to track progress across the 8 setup phases
• Each phase has verification checkboxes that map well to session checkpoints

References

• Conventional Commits Specification
• Semantic Versioning (SemVer)
• semantic-release Documentation
• commitlint Documentation
• Husky Documentation