Cortex-first Snowflake Native App scaffold for Data Quality + Observability parity.

• Native App lifecycle (setup, upgrade, healthcheck)
• Cortex-backed DQ UDF library (no non-Cortex fallback)
• Profiling + classification foundations
• Monitoring + anomaly + incident scaffolding
• Streamlit in Snowflake app shell

• native_app/manifest.yml - Native App package manifest
• native_app/setup.sql - App setup entrypoint
• native_app/scripts/ - Versioned SQL objects and procedures
• native_app/streamlit/ - Streamlit app and page placeholders
• docs/architecture.md - Architecture and dependency notes
• delivery-plan.md - Sprint 1-4 plan with Cortex dependency gates
• tests/sql/smoke_checks.sql - Basic post-deploy checks

1. Create app package from native_app/
2. Run setup.sql in provider packaging workflow
3. Execute scripts in order: 001_init.sql, 002_cortex_udfs.sql, 003_monitoring.sql, 004_remediation.sql
4. Run lifecycle setup and gate checks:
   • CALL APP_CORE.sp_setup_app();
   • CALL APP_CORE.sp_validate_cortex_access();
   • CALL APP_CORE.sp_assert_ready_for_activation();
   • CALL APP_CORE.sp_healthcheck();
5. Validate Sprint 2 rule governance and preview:
   • CALL APP_DQ.sp_preview_rule_result('udf_dq_email_valid', 'user@example.com');
   • CALL APP_DQ.sp_apply_rule_pack('baseline_pack', 'RAW.CUSTOMER');
6. Validate Sprint 3 profiling/monitoring lifecycle:
   • CALL APP_ENGINE.sp_profile_dataset('APP_EVENTS', 'TMP_PROFILE_INPUT');
   • CALL APP_ENGINE.sp_classify_columns_cortex(NULL);
   • CALL APP_ENGINE.sp_inject_synthetic_anomaly('SMOKE_SYNTHETIC', 0.95);
   • CALL APP_ENGINE.sp_run_monitoring_cycle('manual_validation_cycle');
7. Validate Sprint 4 remediation + admin controls:
   • CALL APP_ENGINE.sp_create_task_for_latest_incident('SMOKE_SYNTHETIC', 'Investigate anomaly', 'oncall_user');
   • CALL APP_ENGINE.sp_update_latest_task_for_source('SMOKE_SYNTHETIC', 'IN_PROGRESS', 'Investigation started');
   • CALL APP_CORE.sp_export_diagnostics(24);
   • CALL APP_CORE.sp_enforce_cortex_execution('model-not-allowlisted', 100, 'manual_check', OBJECT_CONSTRUCT());
8. Run tests/sql/smoke_checks.sql

• Release runbook: docs/release-runbook.md
• Rollback checklist: docs/rollback-checklist.md
• Single ordered deployment script: deploy_all.sql (SnowSQL/CLI)

The workflow at .github/workflows/cd-deploy.yml requires these repository Secrets:

• SNOWFLAKE_ACCOUNT
• SNOWFLAKE_USER
• SNOWFLAKE_PASSWORD

Optional repository Variables for environment overrides:

• SNOWFLAKE_DATABASE_STAGING (default: NATIVE_OBS)
• SNOWFLAKE_SCHEMA_STAGING (default: PROD)
• SNOWFLAKE_DATABASE_PROD (default: NATIVE_OBS)
• SNOWFLAKE_SCHEMA_PROD (default: PROD)

If Variables are not set, workflow defaults are used.

This project is now implemented through Sprint 4 baseline and can be extended for production hardening per environment.