$npx -y skills add vaquarkhan/data-engineering-agent-skills --skill microsoft-purview-and-azure-data-governanceGuides agents through Microsoft Purview and Azure-native data governance workflows. Use when designing collections, scans, classifications, lineage, policy boundaries, and governed publishing across ADLS, Synapse, Data Factory, Azure Databricks, Fabric, and Azure analytics estate
| 1 | # Microsoft Purview And Azure Data Governance |
| 2 | |
| 3 | ## Overview |
| 4 | |
| 5 | Use this skill when `Microsoft Purview` is the governance control plane for `Azure` data platforms. It helps agents design metadata collections, classification strategy, scan scope, lineage expectations, and governed publish behavior across Microsoft analytics surfaces. |
| 6 | |
| 7 | ## When to Use |
| 8 | |
| 9 | - designing `Purview` collection and ownership structure |
| 10 | - defining scans, classifications, glossary, and lineage expectations |
| 11 | - governing datasets across `ADLS`, `Synapse`, `Data Factory`, `Azure Databricks`, or `Fabric` |
| 12 | - improving trusted discovery and certification for shared data products |
| 13 | - aligning Azure-native governance with privacy, security, and release controls |
| 14 | |
| 15 | Do not assume `Purview` is only a catalog tool. It often becomes the platform evidence and policy surface for governed analytics. |
| 16 | |
| 17 | ## Workflow |
| 18 | |
| 19 | 1. Define governance scope. |
| 20 | Clarify: |
| 21 | - in-scope platforms |
| 22 | - business domains |
| 23 | - critical data products |
| 24 | - stewardship and ownership model |
| 25 | |
| 26 | 2. Design the metadata operating model. |
| 27 | Decide: |
| 28 | - collections |
| 29 | - glossary boundaries |
| 30 | - classifications and sensitivity labels |
| 31 | - scan cadence and ownership |
| 32 | |
| 33 | 3. Define trusted publish behavior. |
| 34 | Require: |
| 35 | - certification or endorsement rules |
| 36 | - lineage completeness expectations |
| 37 | - ownership visibility |
| 38 | - ties to regulated-data controls where relevant |
| 39 | |
| 40 | 4. Align Azure services with governance. |
| 41 | Check how `Purview` works with `ADLS`, `Synapse`, `Data Factory`, `Databricks`, and `Fabric` rather than treating each system separately. |
| 42 | |
| 43 | 5. Validate operational sustainability. |
| 44 | Make sure scans, classifications, and lineage remain useful as assets, teams, and environments grow. |
| 45 | |
| 46 | ## Common Rationalizations |
| 47 | |
| 48 | | Rationalization | Reality | |
| 49 | | --- | --- | |
| 50 | | "Scanning everything is the same as governing it." | Governance also needs ownership, trust signals, and useful boundaries for consumers. | |
| 51 | | "Purview can be added after pipelines are done." | Late governance usually means weak lineage, poor certification, and inconsistent discovery. | |
| 52 | | "Each Azure service team can manage metadata separately." | Fragmented governance weakens platform-wide trust and policy evidence. | |
| 53 | |
| 54 | ## Red Flags |
| 55 | |
| 56 | - collections do not map to real ownership or domains |
| 57 | - scan scope is broad but lineage and certification are weak |
| 58 | - classifications are inconsistent across Azure services |
| 59 | - `Purview` is disconnected from publish or security decisions |
| 60 | - stewardship expectations depend on tribal knowledge |
| 61 | |
| 62 | ## Verification |
| 63 | |
| 64 | - [ ] Governance scope and stewardship model are explicit |
| 65 | - [ ] Collections, scans, and classifications are intentionally designed |
| 66 | - [ ] Trusted publish behavior includes lineage and certification expectations |
| 67 | - [ ] Azure services align to one governance model |
| 68 | - [ ] The model stays sustainable as adoption grows |