How would you automate forbidden-word checks in CI/CD?

This tests your ability to harden brand voice through automated engineering controls rather than manual policy. A strong answer outlines an AST-aware scanner that distinguishes user-facing strings from internal keys, a CI job that fails builds on violations with configurable severity, and an allowlist or annotation system to suppress false positives. A weak answer relies on manual PR review or naive grep checks that cannot distinguish UI copy from variable names, producing noisy alerts that developers soon ignore.