Data warehouse · 1 skill

Databricks MCP and Claude skills

Pull failed and timed-out job runs, run durations, DBU usage and long-running warehouse queries from Databricks, and report which pipelines broke overnight and why.

Connect Databricks to your agent (MCP)

Databricks managed MCP servers. Databricks hosts MCP servers inside your workspace. Through Unity Gateway they live at https://<workspace-hostname>/ai-gateway/mcp-services/<mcp-name>, and system.ai.dbsql runs SQL on a warehouse with your own Unity Catalog permissions (in beta as of October 2026). None of them covers the Jobs API, but the SQL server can read the system.lakeflow tables. Databricks' coding-agent guides cover connecting Claude Code, Codex and Cursor with OAuth (docs).
Databricks CLI. After databricks auth login --host https://<workspace-hostname>, databricks jobs list-runs --completed-only --output json lists recent runs and databricks jobs get-run-output <run-id> shows a failed task's error.
REST API with a token. The skill reads DATABRICKS_HOST and either DATABRICKS_TOKEN (a personal access token) or DATABRICKS_CLIENT_ID and DATABRICKS_CLIENT_SECRET for an OAuth service principal, which suits scheduled runs.

Run reports as a service principal with CAN VIEW on the jobs, CAN USE on one SQL warehouse and SELECT on the system tables it reads; it can then see every run but can't start, edit or delete a job.

1 Databricks skill

Databricks failed jobs

Every morning: Databricks job runs that failed or timed out since yesterday, with the failing task, the error line and a link, and an ask about repairing.

Data teamsEvery day at 7:30Asks you

What agents do with Databricks

Databricks jobs are read through the Jobs API. GET /api/2.2/jobs/runs/list with completed_only=true, start_time_from and an optional job_id returns runs with their state and result (SUCCESS, FAILED, TIMEDOUT, CANCELED), the state message, start and end times, trigger and run_page_url, and GET /api/2.2/jobs/runs/get-output returns a failed task's error and stack trace. The same history sits in system tables, system.lakeflow.job_run_timeline and system.lakeflow.job_task_run_timeline, which one SQL query can aggregate across every job, and system.billing.usage holds DBU consumption by SKU, workspace and job.

Data teams schedule a morning list of failed and timed-out runs with the error line and a link to each run, jobs whose duration crept up week over week, DBU spend by job or warehouse, and the warehouse queries that ran longest.

databricks.com

Teams that use these

Your Databricks numbers, on one board with everything else

Skills are free to use with any agent. Briefkin is where their results land: one board for every report, number and question your agents send, with a ping when something needs you.

Join the open beta Browse all skills

Free during the open beta. No card.