What problem does it solve? After deploying a data pipeline on Databricks, teams still need to define failure alerts, latency and cost tracking, SLA monitoring, and dashboards by hand. This Skill acts as the final Production Monitoring Agent in a 15-agent SDLC pipeline, turning a Deployment Package into a validated monitoring dashboard and alert rules artifact. ## Core Features & Use Cases - Monitoring Definition Generation: Produces dashboard.json and alert_rules.json covering job/pipeline failure alerts, latency and cost tracking, LLM token usage, data-quality monitoring, and SLA monitoring. - Contract-Validated Output: Returns JSON matching a strict output contract with status, citations, tool calls, and token usage, with one repair retry on validation failure before routing to human review. - Natural-Language Monitoring Queries: Wires a Genie Agents space over monitoring tables so on-call engineers can ask questions like why latency spiked via the Genie Agents Conversation API. - Use Case: A Deployment Agent finishes releasing a pipeline; this agent reads the Deployment Package from the Unity Catalog volume, calls the Foundation Model API to define monitoring, and writes the dashboard and alert artifacts back to the lakehouse. ## Quick Start Hand this agent a Deployment Package for a ticket and ask it to produce the monitoring dashboard and alert rules artifacts for the pipeline.
