What problem does it solve? Manually authoring Databricks Asset Bundles, CI/CD pipeline definitions, and rollback plans for every deployment is error-prone and inconsistent across teams. This Skill acts as Agent 14 of 15 in a Databricks multi-agent SDLC pipeline, turning reviewed and tested code from upstream agents into a validated deployment package. ## Core Features & Use Cases - Bundle Generation: Produces the Databricks Asset Bundle (databricks.yml), CI/CD pipeline definitions for GitHub Actions or Azure DevOps, environment variables, secrets references, approval workflow, and rollback plan. - Contract-Validated Output: Returns JSON matching a strict output contract, runs databricks bundle validate as part of completion, and writes artifacts to /Volumes/{catalog}/state/artifacts/{ticket}/ with state.artifacts and state.agent_runs tracking. - Governed Handoff: Enforces a human approval gate before downstream agents proceed, returns needs_clarification instead of inventing metadata, and routes failures to state.needs_human after one repair retry. - Use Case: After agents 10-12 finish code review and testing for a ticket, invoke this agent to produce a validated deployment package and CI/CD configuration ready for the Production Monitoring Agent. ## Quick Start Ask the agent to generate the Databricks Asset Bundle and CI/CD deployment package for the current ticket using the reviewed code artifacts from agents 10-12.
