ci: bind the branch name to env before writing it to GITHUB_ENV

[kobihikri]

Hi, and thanks for Jina.

In .github/workflows/label-pr.yml, the deploy-to-netlify job writes the branch name into the environment file by interpolation:

- run: |
    echo "BRANCH_NAME=${{ github.head_ref }}" >> $GITHUB_ENV

Actions expands ${{ ... }} into the script text before bash runs, so the branch name becomes part of the command. Git allows $, (, ) and backticks in branch names, and $(...) executes inside double quotes — so a PR from a fork on a branch named x$(id) would run that rather than have its name recorded.

There is a second, quieter effect worth mentioning: because this writes to $GITHUB_ENV, a value containing a newline can define additional environment variables for every later step in the job, not just BRANCH_NAME. The very next step consumes it as ref: ${{ env.BRANCH_NAME }} for a checkout of jina-ai/jina.

The change binds the value to a step-level environment variable first:

- env:
    HEAD_REF: ${{ github.head_ref }}
  run: |
    echo "BRANCH_NAME=$HEAD_REF" >> $GITHUB_ENV

BRANCH_NAME still ends up with the same value and the checkout below is unchanged.

On scope: this is a pull_request trigger, so a fork PR carries a read-only token and no secrets. Hardening rather than a live exploit.

Disclosure: I used AI assistance to help spot this and prepare the change, and I read the job and the step that consumes BRANCH_NAME myself.