skill-freshness-audit

Auditing agent skills for staleness involves fetching embedded documentation URLs to compare skill patterns against live docs. This systematic freshness audit tracks verification anchors and reports drift to identify outdated logic.

Skills drift after a platform release when new API versions or deprecated features alter official documentation. Running a freshness audit checks skill logic against the latest Databricks and MLflow docs to detect this drift.

Checking upstream sync requires tracking lineage to databricks-solutions/ai-dev-kit. An upstream sync audit compares your skills against the source repository to identify unmerged changes and maintain dependency currency.

A skill freshness check for volatility classification is needed when verifying skills that change frequently. It identifies skills overdue for verification based on their volatility and reports staleness before major updates.

Verifying agent skills against live docs uses automated verification anchors. The audit fetches official documentation URLs embedded in skills and compares patterns against live sources to report discrepancies without manual checks.

Skill freshness auditing works for standalone skills by focusing on embedded documentation URLs and staleness detection. Even without external dependencies, it classifies volatility and checks skill logic against live docs.