Databricks Serverless Migration

databricks/databricks-agent-skillsOfficial

If you're moving Databricks workloads from classic clusters to serverless compute, this walks you through the compatibility minefield. It scans your notebooks for RDD usage, DBFS paths, Hive metastore references, streaming triggers, and other patterns that break on the serverless Spark Connect architecture, then gives you concrete DataFrame API rewrites and config changes. The four-step workflow (ingest, analyze, test, validate) is methodical, and the decision tree separating temporary limitations from actual blockers saves you from debugging things Databricks hasn't implemented yet. Worth noting this is serverless-specific migration, not general DBR version upgrades. If you're still figuring out whether to move at all, the category breakdown helps you estimate effort before committing.

skills.sh installs: 681
GitHub stars: 393
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add databricks/databricks-agent-skills --skill databricks-serverless-migration --agent claude-code

Installs into .claude/skills of the current project.

Files

Serverless Compute Migration

FIRST: Use the parent databricks-core skill for CLI basics, authentication, and profile selection.

Analyze existing Databricks code for serverless compute compatibility and guide migration from classic clusters. The skill follows a 4-step migration lifecycle: Ingest the workload → Analyze for compatibility → Test via A/B comparison → Validate and iterate.

Prefer the Compute Migrations API for Jobs

To migrate a job, first check whether databricks-serverless-migration-api is installed and its prerequisites are available. If so, use it to start and manage the migration. The service backs up the job, builds and optionally runs a serverless working copy, and changes the original only after review. Use the manual workflow below when:

the API skill, databricks compute-migrations CLI commands, or Compute Agent preview are not available,
the workload is a standalone notebook or pipeline,
the service reports the job as not supported, or
the user wants you to make the changes yourself.