Claude Skill

databricks

Operate Databricks bundles, jobs, pipelines, and Unity Catalog.

LLM Mart · 0 points · 1 views 0 listing impressions 0 install-command copies
Download fmind-dot-skills_databricks-3bbe830.zip · 1 KB

Install

skills CLI npx skills add https://github.com/fmind/dot/tree/main/skills/databricks
Claude Code claude plugin marketplace add https://llmmart.ai/marketplace.json && claude plugin install fmind-dot@llmmart
Git git clone https://github.com/fmind/dot.git

The skills CLI installs just this skill, for any of its supported agents. Claude Code installs the whole fmind/dot collection as a plugin from our marketplace. Git is the plain clone.

Skill manifest

Databricks CLI

Operate Databricks workspaces, Declarative Automation Bundles (DABs), clusters, jobs, Lakeflow pipelines, and Unity Catalog with databricks.

Confirm workspace, --profile <profile>, and --target before mutations or billable runs. Env vars like DATABRICKS_TOKEN bypass host-profile matching and send tokens to whatever host a bundle names—set DATABRICKS_BUNDLE_STRICT_PROFILE_MATCH=1 on untrusted bundles (requires ~/.databrickscfg, so omit in CI).

Workflow

  1. Check profile and caller (prefer OAuth U2M databricks auth login --host <workspace-url> --profile <profile> over static PATs):

    databricks auth profiles --skip-validate
    databricks current-user me --profile <profile>
    
  2. Validate, plan, and deploy bundles (re-plan after any local edit since deploy --plan detects remote drift only, and gitignore plan.json so bundle sync skips it):

    databricks bundle validate --target dev --profile <profile>
    databricks bundle plan --target dev --profile <profile> -o json > plan.json
    databricks bundle deploy --target dev --profile <profile> --plan plan.json
    databricks bundle run <job-or-pipeline-key> --target dev --profile <profile>
    
  3. Inspect Unity Catalog, compute, and runs (pass --limit on list commands, supported in CLI 1.25.0, to avoid full-catalog scans):

    databricks catalogs get <catalog> --profile <profile> --output json
    databricks schemas get <catalog>.<schema> --profile <profile> --output json
    databricks tables get <catalog>.<schema>.<table> --profile <profile> --output json
    databricks clusters get <cluster-id> --profile <profile> --output json
    databricks jobs list-runs --job-id <job-id> --limit 10 --profile <profile>
    
  4. Authorize consequential changes: --target prod, cluster restarts, ACLs, and schema DDL require user approval.

Gotchas

  • Always pass --target: omitting it hits the default target or errors.
  • Direct bundle engine since CLI 1.20: bundle.engine: terraform errors and first deploy migrates existing Terraform state (pin CLI 1.19.x if Terraform is still required; keep local and CI CLI versions aligned).
  • Keyless CI first: in GitHub Actions use workload identity federation (DATABRICKS_AUTH_TYPE: github-oidc, DATABRICKS_HOST, DATABRICKS_CLIENT_ID, id-token: write), else OAuth M2M (DATABRICKS_CLIENT_ID + DATABRICKS_CLIENT_SECRET), and env-only DATABRICKS_TOKEN as a last resort—never commit tokens.
  • Ignore databricks aitools install stderr hints (emitted under Claude Code in CLI 1.23.0 with no opt-out flag): preview skills into a non-discovery <dir> via databricks aitools install --path <dir> --skills <name> and follow vendor-skill policy.
  • Check cluster autotermination before starting compute to prevent idle spend.

Official Skills

  • Upstream: databricks/databricks-agent-skills (select from top-level skills/, not plugins/databricks/).

Documentation

Files (dot)
  • SKILL.md 3.8 KB
    ---
    name: databricks
    description: "Operate Databricks bundles, jobs, pipelines, and Unity Catalog."
    license: MIT
    metadata:
      kind: connector
      author: Médéric HURIER (Fmind)
      source: github.com/fmind/dot/tree/main/skills/databricks
      created: "2026-09-16"
      updated: "2026-11-10"
    ---
    
    # Databricks CLI
    
    Operate Databricks workspaces, Declarative Automation Bundles (DABs), clusters, jobs, Lakeflow pipelines, and Unity Catalog with `databricks`.
    
    Confirm workspace, `--profile <profile>`, and `--target` before mutations or billable runs. Env vars like `DATABRICKS_TOKEN` bypass host-profile matching and send tokens to whatever host a bundle names—set `DATABRICKS_BUNDLE_STRICT_PROFILE_MATCH=1` on untrusted bundles (requires `~/.databrickscfg`, so omit in CI).
    
    ## Workflow
    
    1. **Check profile and caller** (prefer OAuth U2M `databricks auth login --host <workspace-url> --profile <profile>` over static PATs):
    
       ```bash
       databricks auth profiles --skip-validate
       databricks current-user me --profile <profile>
       ```
    
    1. **Validate, plan, and deploy bundles** (re-plan after any local edit since `deploy --plan` detects remote drift only, and gitignore `plan.json` so bundle sync skips it):
    
       ```bash
       databricks bundle validate --target dev --profile <profile>
       databricks bundle plan --target dev --profile <profile> -o json > plan.json
       databricks bundle deploy --target dev --profile <profile> --plan plan.json
       databricks bundle run <job-or-pipeline-key> --target dev --profile <profile>
       ```
    
    1. **Inspect Unity Catalog, compute, and runs** (pass `--limit` on `list` commands, supported in CLI 1.25.0, to avoid full-catalog scans):
    
       ```bash
       databricks catalogs get <catalog> --profile <profile> --output json
       databricks schemas get <catalog>.<schema> --profile <profile> --output json
       databricks tables get <catalog>.<schema>.<table> --profile <profile> --output json
       databricks clusters get <cluster-id> --profile <profile> --output json
       databricks jobs list-runs --job-id <job-id> --limit 10 --profile <profile>
       ```
    
    1. **Authorize consequential changes**: `--target prod`, cluster restarts, ACLs, and schema DDL require user approval.
    
    ## Gotchas
    
    - **Always pass `--target`**: omitting it hits the default target or errors.
    - **Direct bundle engine since CLI 1.20**: `bundle.engine: terraform` errors and first deploy migrates existing Terraform state (pin CLI 1.19.x if Terraform is still required; keep local and CI CLI versions aligned).
    - **Keyless CI first**: in GitHub Actions use [workload identity federation](https://docs.databricks.com/aws/en/dev-tools/auth/provider-github) (`DATABRICKS_AUTH_TYPE: github-oidc`, `DATABRICKS_HOST`, `DATABRICKS_CLIENT_ID`, `id-token: write`), else OAuth M2M (`DATABRICKS_CLIENT_ID` + `DATABRICKS_CLIENT_SECRET`), and env-only `DATABRICKS_TOKEN` as a last resort—never commit tokens.
    - **Ignore `databricks aitools install` stderr hints** (emitted under Claude Code in CLI 1.23.0 with no opt-out flag): preview skills into a non-discovery `<dir>` via `databricks aitools install --path <dir> --skills <name>` and follow [vendor-skill policy](../agent-project/references/vendor-skills.md).
    - **Check cluster autotermination** before starting compute to prevent idle spend.
    
    ## Official Skills
    
    - Upstream: `databricks/databricks-agent-skills` (select from top-level `skills/`, not `plugins/databricks/`).
    
    ## Documentation
    
    - [Databricks CLI guide](https://docs.databricks.com/aws/en/dev-tools/cli) · [Declarative Automation Bundles](https://docs.databricks.com/aws/en/dev-tools/bundles)
    - [Unity Catalog guide](https://docs.databricks.com/aws/en/data-governance/unity-catalog)
    - Releases: [Databricks CLI Releases](https://github.com/databricks/cli/releases)
    - Companion skills: [duckdb](../duckdb/SKILL.md) (local data), [python-stack](../python-stack/SKILL.md) (Python development).
    

Comments (0)

Sign in to join the conversation.

No comments yet.

Reviews (0)

No reviews yet.