fix(ci): Prevent script injection in ai-pr-review workflow

[ayushiahjolia]

Issue #, if available: N/A

Description of changes:
Mitigate potential script injection in the ai-pr-review.yml workflow by assigning github.event.pull_request.head.sha to an environment variable instead of interpolating it directly in a run: block.

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

[github-actions]

Codex AI review

No actionable findings. Residual risk is limited to the workflow change not being runtime-tested in this review.

Reviewed commit 4c6579f92a99a543def7e37ad57049beab25c467. Workflow run

[github-actions]

Claude AI review

This PR mitigates a potential script-injection vector in .github/workflows/ai-pr-review.yml by moving github.event.pull_request.head.sha from direct ${{ }} interpolation inside the run: shell into an env: variable (HEAD_SHA) and referencing it as $HEAD_SHA. This is the correct, recommended GitHub Actions pattern: the shell no longer evaluates untrusted expression-substituted text. The variable name is consistent between env: and the run: step, the value is properly quoted, and the echoed message semantics are unchanged. No correctness, typing, error-handling, or lifecycle issues were found. Residual test risk: GitHub Actions workflows are not unit-tested in this repo, so this defensive, no-functional-change hardening relies on review rather than automated verification; risk is minimal.

Reviewed commit 4c6579f92a99a543def7e37ad57049beab25c467. Workflow run

[zhongkechen]

is sha field injectable?