data-governance-mcp

> User in Cursor:
"@data-governance I need to write a predictive anomaly detection pipeline for furnace_telemetry.csv,
but corporate compliance prohibits uploading internal telemetry and customer data to external LLMs."

> Claude (via Data Governance MCP):
"I have audited `furnace_telemetry.csv` and secured your development session:

1. 🚨 PII Sanitization: Flagged unmasked email addresses and API keys in column `operator_email`.
2. 🧪 Synthetic Digital Twin: Generated `synthetic_twin.csv` (500 rows) matching exact statistical
   moments (μ=1,245.8°C, σ=42.1°C), categorical distributions, and null ratios with differential privacy (ε=1.0).
3. ⚡ Token Optimization: Compressed 14,250 tokens down to 680 tokens (-95.2% prompt reduction).
4. 🚀 Production Artifacts:
   - Generated `schema.yml` with dbt tests (`not_null`, `unique`, and Tukey outlier range tests).
   - Exported interactive executive audit report to `reports/audit_dashboard.html`.

You can now develop and test your predictive model in Cursor using the synthetic twin with zero compliance risk."

data-governance-mcp dashboard telemetry.csv --out audit_report.html

data-governance-mcp audit examples/sample_datasets/industrial_furnace_telemetry.csv

┌─────────────────── DATA GOVERNANCE & PRIVACY TWIN AUDIT ────────────────────┐
│ Dataset: Delimited file (industrial_furnace_telemetry.csv)                  │
│ Records: 13  |  DAMA-DMBOK Score: 94.6 / 100 (EXCELLENT)                    │
│ Security & PII Risk: HIGH (2 findings flagged)                              │
└─────────────────────────────────────────────────────────────────────────────┘
                     DAMA-DMBOK 6 Core Quality Dimensions                      
┌──────────────┬────────┬────────┬─────────────┬──────────────────────────────┐
│ Dimension    │ Weight │  Score │   Status    │ Diagnostics                  │
├──────────────┼────────┼────────┼─────────────┼──────────────────────────────┤
│ Completeness │  22%   │  97.8% │  [ PASS ]   │ 2 null values detected (2.2% │
│              │        │        │             │ missingness).                │
│ Uniqueness   │  18%   │  84.6% │  [ FAIL ]   │ 1 exact duplicate row found. │
│ Validity     │  22%   │ 100.0% │  [ PASS ]   │ All columns conform to types.│
│ Accuracy     │  16%   │  89.5% │ [ WARNING ] │ 1 statistical outlier (IQR). │
│ Consistency  │  12%   │ 100.0% │  [ PASS ]   │ No logical contradictions.   │
│ Timeliness   │  10%   │  95.0% │  [ PASS ]   │ Evaluated on 'timestamp'.    │
└──────────────┴────────┴────────┴─────────────┴──────────────────────────────┘
┌──────────────────────────────── Action Plan ────────────────────────────────┐
│ Recommended Remediation Workflow:                                           │
│ 1. Generate privacy twin:  data-governance-mcp twin telemetry.csv           │
│ 2. Export dbt tests:       data-governance-mcp dbt telemetry.csv            │
│ 3. Generate HTML report:   data-governance-mcp dashboard telemetry.csv      │
└─────────────────────────────────────────────────────────────────────────────┘

# Using uv (Recommended)
uv tool install data-governance-mcp

# Or via standard pip
pip install data-governance-mcp

{
  "mcpServers": {
    "data-governance": {
      "command": "python",
      "args": ["-m", "data_governance_mcp.server"]
    }
  }
}

pytest -v

tests/test_dimensions.py::test_completeness_perfect PASSED               [  5%]
tests/test_dimensions.py::test_completeness_with_nulls_and_blanks PASSED [ 10%]
tests/test_dimensions.py::test_uniqueness_with_duplicates PASSED         [ 15%]
tests/test_dimensions.py::test_accuracy_outliers PASSED                  [ 21%]
tests/test_dimensions.py::test_consistency_chronology_inversion PASSED   [ 26%]
tests/test_dimensions.py::test_evaluate_all_summary PASSED               [ 31%]
tests/test_loader.py::test_load_inline_csv PASSED                        [ 36%]
tests/test_loader.py::test_load_sqlite PASSED                            [ 42%]
tests/test_pii_detector.py::test_pii_detection_clean PASSED              [ 47%]
tests/test_pii_detector.py::test_pii_detection_email_and_secrets PASSED  [ 52%]
tests/test_server.py::test_audit_dataset_tool PASSED                     [ 57%]
tests/test_server.py::test_profile_schema_tool PASSED                    [ 63%]
tests/test_server.py::test_detect_pii_tool PASSED                        [ 68%]
tests/test_server.py::test_evaluate_dama_dimensions_tool PASSED          [ 73%]
tests/test_server.py::test_suggest_remediations_tool PASSED              [ 78%]
tests/test_wow_features.py::test_synthetic_twin_generator PASSED         [ 84%]
tests/test_wow_features.py::test_token_compressor PASSED                 [ 89%]
tests/test_wow_features.py::test_dbt_exporter PASSED                     [ 94%]
tests/test_wow_features.py::test_dashboard_exporter PASSED               [100%]

============================= 19 passed in 2.19s ==============================