databricks-solutions/genie-code-skills-demo
Developer toolsこのページは AI により自動翻訳されています。技術的な詳細は原文をご確認ください。 原文を見る
Example Genie Code skills, custom instructions, and MCP setup for enforcing enterprise data engineering standards across Databricks development.
概要
Example Genie Code skills, custom instructions, and MCP setup for enforcing enterprise data engineering standards across Databricks development. Genie Code is Databricks' AI coding assistant. It can: - -- streaming tables, materialized views, CDC, medallion architecture - -- Python, SQL, Scala, R - -- MLflow experiment tracking, model training, feature engineering - -- data visualizations and business intelligence - -- full-stack applications on the lakehouse - -- governance, permissions, lineage - -- scheduling, dependencies, monitoring Genie Code supports (task-specific instructions following the open Agent Skills standard -- each skill is a folder with a SKILL.md), (user- and workspace-level, plus auto-discovered AGENTS.md / CLAUDE.md files), and (Model Context Protocol) connections that fetch enterprise standards from external sources like GitHub. This repo provides example skills, instructions, and tooling organized by domain. Each phase adds a new area of coverage.
README
以下は原文 README の抜粋です。原文を見る。
Genie Code Skills Demo
Example Genie Code skills, custom instructions, and MCP setup for enforcing enterprise data engineering standards across Databricks development.
What is Genie Code?
Genie Code is Databricks’ AI coding assistant. It can:
- Author SDP pipelines – streaming tables, materialized views, CDC, medallion architecture
- Develop notebooks – Python, SQL, Scala, R
- Build DSML workflows – MLflow experiment tracking, model training, feature engineering
- Create AI/BI dashboards – data visualizations and business intelligence
- Develop Databricks Apps – full-stack applications on the lakehouse
- Manage Unity Catalog – governance, permissions, lineage
- Orchestrate jobs and workflows – scheduling, dependencies, monitoring
Genie Code supports skills (task-specific instructions following the open Agent Skills standard – each skill is a folder with a SKILL.md), instructions (user- and workspace-level, plus auto-discovered AGENTS.md / CLAUDE.md files), and MCP (Model Context Protocol) connections that fetch enterprise standards from external sources like GitHub.
Demo Coverage
This repo provides example skills, instructions, and tooling organized by domain. Each phase adds a new area of coverage.
| Phase | Date | Domain | What’s Included |
|---|---|---|---|
| Phase 1 | April 2026 | Data Engineering | SDP pipeline skills, PII management, table governance, MCP-based standards enforcement, sample financial data generation |
| Phase 2 | May 2026 | DSML | sentiment-analysis skill, AI-function patterns (ai_analyze_sentiment, ai_classify, ai_extract), AI/BI Bakehouse Marketplace install |
| Planned | – | Dashboards, Governance | Additional skill domains and MCP integrations |
What This Repo Contains
| Folder | Contents |
|---|---|
skills/ |
One folder per skill (SKILL.md + templates). Deploy with a vibe tool, ./skills/deploy.sh, or a manual copy; also served via MCP |
AGENTS.md |
Auto-discovered enterprise standards – Genie Code reads this automatically from the workspace directory tree |
instructions/ |
Instruction templates (placeholders). Copy and fill in for your workspace – by hand or with a vibe tool. Do not treat these as ready-to-upload. |
mcp/ |
Unity Catalog HTTP MCP connection to GitHub (no-scope PAT). See mcp/README.md. |
sample_data_gen/ |
Synthetic financial data generation notebook (uses dbldatagen) |
marketplace_data/ |
Installer notebook + config template for sourcing datasets from the Databricks Marketplace (currently AI/BI Bakehouse) |
推奨ツール
別のキーワードを試すか、フィルタを外してください。
インストール
npx skillfish add databricks-solutions/genie-code-skills-demo