Pipeline security and supply chain

Least-privilege tokens, pinning actions by SHA, secret scanning, dependency and container scanning, provenance, and running untrusted pull request code safely.

Least privilege and pinned dependencies

# the token is read-only unless a job asks for more
permissions:
  contents: read

jobs:
  test:
    runs-on: ubuntu-latest
    permissions:
      contents: read              # explicit, so a later default cannot widen it
    steps:
      # pin by full commit SHA: a tag can be moved to point at new code
      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683  # v4.2.2
      - uses: actions/setup-node@39370e3970a6d050c480ffad4ff0ed4d3fdee5af  # v4.1.0
        with: { node-version: 22, cache: npm }
      - run: npm ci --ignore-scripts
      - run: npm test

  publish:
    needs: test
    runs-on: ubuntu-latest
    permissions:
      contents: read
      packages: write
      id-token: write             # only this job can sign
    steps:
      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683
      - run: ./publish.sh

• A tag is a mutable pointer. Pinning a third-party action to a commit SHA is the only way to know what code will run next month.
• npm ci --ignore-scripts stops a dependency from running post-install code, which is a common supply-chain attack path.
• Grant write permissions only to the job that genuinely needs them — publishing, releasing or commenting.
• Never echo a secret. Masked values in logs can still leak through base64, a URL or a diff, so scan the full output.

Running untrusted code

# untrusted workflow: no secrets, read-only token, no writes
on:
  pull_request_target:            # runs in the base repo, so be careful

jobs:
  label:
    runs-on: ubuntu-latest
    permissions:
      contents: read
      pull-requests: write
    steps:
      # NEVER check out the pull request head here: it would run untrusted code
      # with a write token. Comment on metadata only.
      - uses: actions/github-script@v7
        with:
          script: |
            await github.rest.issues.addLabels({
              ...context.repo, issue_number: context.issue.number, labels: ['needs-review']
            });

Scanning and provenance

jobs:
  scan:
    runs-on: ubuntu-latest
    permissions:
      contents: read
      security-events: write
    steps:
      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683
        with: { fetch-depth: 0 }

      - name: Secret scan
        uses: gitleaks/gitleaks-action@v2

      - name: Dependency review
        if: github.event_name == 'pull_request'
        uses: actions/dependency-review-action@v4
        with:
          fail-on-severity: high

      - name: Container scan
        uses: aquasecurity/trivy-action@0.28.0
        with:
          image-ref: ghcr.io/example/app:sha-${{ github.sha }}
          severity: CRITICAL,HIGH
          exit-code: '1'

      - name: Attest build provenance
        uses: actions/attest-build-provenance@v1
        with:
          subject-name: ghcr.io/example/app
          subject-digest: ${{ steps.build.outputs.digest }}
          push-to-registry: true

Scanning is only useful with a rule for what happens on a finding. Decide in advance which severities fail the build, which open a ticket, and which are accepted with a documented reason — otherwise the output becomes noise nobody reads.

FAQ

Related

Building and publishing container images in CI Infrastructure as code in the pipeline

Last refreshed 2026-09-18.