Introducing linting to an established design token system often fails due to poor sequencing rather than technical issues. A phased rollout is recommended: start with a baseline audit in report-only mode, fix critical errors before enforcing any rules, then gradually enable blocking checks in pre-commit and CI. Non-critical warnings should be treated as backlog and addressed incrementally. Ownership must be clearly defined — someone specific needs to own lint rule configuration, exception approvals, and documentation. Exceptions should be tracked with scope, rationale, approver, and expiry. Success is measured by metrics like time-to-fix linting errors, critical CI failures per week, exception request trends, and reduction in production token incidents. The key insight is that gradual, predictable rollouts with team buy-in outperform strict enforcement from day one.
