ARS.FLOW
← Back to skills
ETL & Data ValidationHigh ReliabilityHigh-Risk Workflow

ETL Reconciliation Check

Compares source and target ETL summaries with deterministic checks to catch data-loss and transformation drift before release.

Version

0.1.0

Last Updated

Apr 22, 2026

Verification Type

reconciliation checks, integration tests

Downloads

0

Required inputs

  • source_summary (json)

    Source dataset counts, duplicates, and checksums.

  • target_summary (json)

    Target dataset counts, duplicates, and checksums.

Expected outputs

  • reconciliation_report (markdown)

    Deterministic report of count and checksum mismatches.

Included checks and assets

  • scripts/reconcile_summaries.py (script)

    Compares source and target summary JSON metrics.

  • references/reconciliation-policy.md (reference)

    Policy thresholds and blocking criteria for reconciliation checks.

  • references/window-alignment-guide.md (reference)

    Guide to prevent false mismatches from window skew.

  • references/mismatch-triage-playbook.md (reference)

    Step-by-step triage flow for reconciliation mismatches.

Failure modes

  • Summary files exclude key business dimensions.
  • Tolerance configuration is too permissive.
  • Mismatch triage is delayed and blocks release timelines.

Ideal use cases

  • Warehouse ingestion validation
  • Migration data cutovers
  • Batch data quality gates

Example runs

Daily orders ETL reconciliation

Validated sample run

Flags checksum mismatch despite matching row counts.

Input preview

source_summary.json + target_summary.json

Output preview

FAIL with checksum discrepancy on order_total

Changelog summary

  • 0.1.0 · Apr 22, 2026

    Initial release for deterministic ETL reconciliation.

Links

Inspect the source, read authored documentation, or download the published skill bundle.

DownloadSourceDocs