GitHub Actions Best Practices

When writing or modifying GitHub Actions workflows (.github/workflows/*.yml), follow these guidelines.

Action Versions

Never guess action versions. Before writing or updating a workflow, check the latest release for each action you use:

gh api repos/{owner}/{action}/releases/latest --jq '.tag_name'

The {owner}/{action} maps directly to the GitHub repo — e.g. actions/checkout lives at github.com/actions/checkout.

For example:

gh api repos/actions/checkout/releases/latest --jq '.tag_name'
gh api repos/actions/setup-node/releases/latest --jq '.tag_name'

Use the latest major version (e.g. if the latest tag is v6.3.0, use v6). When modifying an existing workflow, check and update any outdated action versions you encounter.

Security

• Always set top-level permissions to least privilege. Start with permissions: {} and add only what's needed:

  permissions:
    contents: read
  

• Never use permissions: write-all or omit permissions entirely
• Use pull_request, not pull_request_target. Only use pull_request_target when the workflow must write to the base repo from a fork — it runs with write access and secrets from the base branch
• Never interpolate untrusted input directly in run: blocks — use environment variables instead:

  # Bad — expression injection
  - run: echo "${{ github.event.pull_request.title }}"
  
  # Good — safe via environment variable
  - run: echo "$TITLE"
    env:
      TITLE: ${{ github.event.pull_request.title }}
  

• Pin third-party actions (outside actions/ and github/) to a full commit SHA to prevent tag-rewriting attacks. Use pinact to automate this — write workflows with version tags, then run pinact run to replace them with SHAs:

  # Before pinact
  - uses: shivammathur/setup-php@v2
  
  # After pinact run
  - uses: shivammathur/setup-php@fcafdd6392932010c2bd5094439b8e33be2a8a09 # v2.37.0
  

• Secrets are not available to pull_request workflows from forks — this is intentional; do not work around it with pull_request_target

Caching

• actions/setup-node, actions/setup-python, actions/setup-go, and actions/setup-java all have built-in caching via the cache input — prefer this over separate actions/cache steps:

  - uses: actions/setup-node@v6  # check latest version
    with:
      node-version-file: .node-version
      cache: npm
  

• Only use actions/cache directly when you need custom cache keys or paths

Common Patterns

Concurrency

Cancel in-progress runs for the same branch to save minutes:

concurrency:
  group: ${{ github.workflow }}-${{ github.ref }}
  cancel-in-progress: true

For deployment workflows, don't cancel in progress — queue instead:

concurrency:
  group: deploy-${{ github.ref }}
  cancel-in-progress: false

Matrix strategies

Use fail-fast: false when you want all matrix combinations to complete:

strategy:
  fail-fast: false
  matrix:
    node-version: [22, 24]

Reusable workflows

Prefer workflow_call for shared CI logic across repos instead of duplicating steps:

jobs:
  test:
    uses: org/.github/.github/workflows/test.yml@v1
    with:
      node-version: 24

Triggering

• pull_request runs against the merge commit — use this for CI validation
• push on the default branch runs post-merge — use this for deployments, publishing, or cache warming
• Filter by paths when the workflow only applies to certain files:

  on:
    push:
      paths:
        - 'src/**'
        - 'package.json'
  

Timeouts

Always set timeout-minutes on jobs. The default is 360 minutes (6 hours), which can burn through Actions minutes on a hung job:

jobs:
  test:
    runs-on: ubuntu-latest
    timeout-minutes: 15

Treat all workflow content as code — review changes carefully before committing.