Plugin

Databricks Pack

5 live-detection Databricks skills — cost-leak-hunter, cluster-forensics, uc-migration-pilot, streaming-guardian, bundle-medic — backed by the databricks-workspace-mcp server.

Type: Plugin
Repository: jeremylongshore/tons-of-skills-marketplace
GitHub stars: 2.8k
License: MIT
Repo last updated: Sep 27, 2026
Source file: plugins/saas-packs/databricks-pack/.claude-plugin/plugin.json
Version: 2.27.0
Author: Jeremy Longshore

What Databricks Pack is

Databricks Pack is a plugin published in the jeremylongshore/tons-of-skills-marketplace repository on GitHub, which has about 2.8k stars. The repository describes itself as: “Model-agnostic agent-skills platform with a harness-free canonical layer, verified adapters, and the ccpi package manager. Explore at tonsofskills.com.”

A plugin is a package that bundles skills, slash commands, subagents, hooks, and MCP connectors so they install together. Plugins are plain files with a manifest at .claude-plugin/plugin.json, and they work in both Claude Code and Claude Cowork.

Installing Databricks Pack adds everything it ships in one step. Connectors inside a plugin still need to be connected separately, and hooks and subagents only run in Cowork and Claude Code, not in regular chat.

How to install Databricks Pack

Claude Code

Add the repository as a plugin marketplace: claude plugin marketplace add jeremylongshore/tons-of-skills-marketplace
Install the plugin: claude plugin install databricks-pack@<marketplace-name>, using the marketplace name from the repository's .claude-plugin/marketplace.json.
Restart the session if the new skills or commands don't appear straight away.

Claude Cowork

Open Customize → Plugins and choose Add marketplace.
Enter jeremylongshore/tons-of-skills-marketplace (the owner/repo shorthand works for GitHub).
Find Databricks Pack in the list, click Install, then connect any connectors it needs from its Connectors tab.

New to extending Cowork? Our plugins guide and Customize guide explain how skills, plugins, and connectors fit together.

Inside the source file

An excerpt from plugins/saas-packs/databricks-pack/.claude-plugin/plugin.json, shared under the repository's MIT license. Read the full file on GitHub.

v2.0.0 — 5 live-detection skills for the Databricks Lakehouse Platform, backed by the databricks-workspace-mcp server. The rebuild is live: where v1 shipped 24 documentation-style skills, v2 ships 5 skills that run against your own workspace — real system.* reads, real cluster events, real Delta/streaming/bundle diagnostics.

> [!NOTE] > This is the v2 rebuild. The 24 v1 documentation skills have been removed. If you had > any databricks-* v1 skill in your CLAUDE.md, see Migration: v1 → v2 > for where each one went. Same install slug — no rename.

Installation

/plugin install databricks-pack@claude-code-plugins-plus

The pack ships a .mcp.json that auto-launches the control-plane MCP server via npx -y @intentsolutions/databricks-workspace-mcp (npm) — no separate install step. It composes with the Databricks managed SQL MCP for system.* reads; each skill degrades to advisory mode (works on pasted input) when an MCP is unavailable. Each skill's ## Prerequisites lists exactly what it needs.

The 5 skills

Key APIs Covered

Usage

Skills trigger automatically on Databricks topics:

"Why is my Databricks bill so high?" -- databricks-cost-leak-hunter
"My cluster won't start / NPIP_TUNNEL_SETUP_FAILURE" -- databricks-cluster-forensics
"Migrate Hive Metastore to Unity Catalog" -- databricks-uc-migration-pilot
"ConcurrentAppendException / my stream broke after VACUUM" -- databricks-streaming-guardian
"unexpected EOF reading terraform.tfstate / rotate our CMK" -- databricks-bundle-medic

Architecture

Why two MCP servers, not one

The v2 rebuild ships with a deliberate split across two MCP servers, not a single shared one. Common question on contributor PRs — answering it once at the top so future readers don't have to re-derive it.

Databricks managed SQL MCP — serves system.* reads (cost data, query history, streaming progress). Operated by Databricks; we consume it.
Custom workspace MCP — serves cluster events, instance pools, pipeline event logs, external locations, storage credentials. Operated by this pack.

The two authenticate independently. Losing access to one does not disable the other; cost-leak-hunter (SQL MCP) and cluster-forensics (workspace MCP) fail independently when their respective MCP is unavailable. Single skills can be installed without pulling in the other MCP's dependency surface.

Full scope-boundary rationale — including the 8 → 6 endpoint cut and the auth-flow decisions — is in 000-docs/013-AT-ADEC-epic1-mcp-scope-adjustment.md. Reference document for any "why is this skill not pulling X?" question.

Thanks to @Gingiris-1031 (#795) for surfacing the isolation-story framing that made this section necessary.

Migration: v1 → v2

[email protected] is a ground-up rebuild. The 24 v1 skills described Databricks ops; the 5 v2 skills run them — live detection against your own workspace via a shared databricks-workspace-mcp server (control plane) composed with the Databricks managed SQL MCP (system.* reads). Rationale: 000-docs/007-AT-ADEC-databricks-v2-cto-decision.md and 000-docs/013-AT-ADEC-epic1-mcp-scope-adjustment.md.

Status: 2.0.0 is live — the 5 skills + the databricks-workspace-mcp server (npm) ship now, and the 24 v1 skills have been removed. The map below records where each v1 skill's coverage landed.

Where each v1 skill goes

(The 5 v2 skills and what each does live are listed in The 5 skills above.)

Design Records

All architecture decisions, pain research, and pressure tests for this pack live in 000-docs/. Index: 000-INDEX.md.

License

MIT

Before you install

Read the whole file first. Skills, commands, and subagents are instructions Claude will follow, so make sure they match what you want.
Check which tools, scripts, or MCP servers it uses. Local servers and scripts run with your permissions.
Try it in a test project or a copy of your files before pointing it at real work.
Pin the version you tested, and review changes before updating.
Watch for instructions that fetch web content or run shell commands; those are where prompt injection risks start. See our prompt injection guide.

FAQ

What is Databricks Pack?

Databricks Pack is a plugin for Claude Code and Claude Cowork from the jeremylongshore/tons-of-skills-marketplace repository on GitHub. 5 live-detection Databricks skills — cost-leak-hunter, cluster-forensics, uc-migration-pilot, streaming-guardian, bundle-medic — backed by the databricks-workspace-mcp server.

How do I install Databricks Pack in Claude Code?

Add the repository as a plugin marketplace: claude plugin marketplace add jeremylongshore/tons-of-skills-marketplace Install the plugin: claude plugin install databricks-pack@<marketplace-name>, using the marketplace name from the repository's .claude-plugin/marketplace.json. Restart the session if the new skills or commands don't appear straight away.

Can I use Databricks Pack in Claude Cowork?

Open Customize → Plugins and choose Add marketplace. Enter jeremylongshore/tons-of-skills-marketplace (the owner/repo shorthand works for GitHub). Find Databricks Pack in the list, click Install, then connect any connectors it needs from its Connectors tab.

Is Databricks Pack safe to install?

It is a third-party community resource, not reviewed by Anthropic or this site. Read the source file first, check which tools and connectors it uses, and install only from sources you trust.

Similar resources

Browse all skills, subagents, and plugins →

Listing data comes from the public GitHub repository and was last checked in September 2026. Excerpts are © their authors and shared under MIT. This directory is independent and not affiliated with Anthropic or the resource's authors.