sian-data-correction-policy

Determines correct remediation paths for data errors in BigQuery Raw, Silver, and Gold layers.

Updated May 21, 2026
One-click install
npx skills add https://github.com/gestaosian/gtsi-claude-skills --skill sian-data-correction-policy-gestaosian
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: sian-data-correction-policy
Source: https://github.com/gestaosian/gtsi-claude-skills/tree/main/plugins/gtsi-ops-plugin/skills/sian-data-correction-policy
Command: npx skills add https://github.com/gestaosian/gtsi-claude-skills --skill sian-data-correction-policy-gestaosian

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Data engineers and analysts on the SIAN platform frequently receive requests to patch data directly in BigQuery tables, which breaks pipeline reproducibility and creates divergence from source systems. This Skill enforces the architectural rule that no direct data corrections are allowed in any derived layer and redirects every request to the proper fix. ## Core Features & Use Cases - Absolute correction policy: Blocks direct SQL updates, Python scripts, notebooks, correction DAGs, and dbt overrides against Raw, Silver, and Gold layers. - Remediation decision table: Maps each error location (source system, Raw extraction bug, Silver/Gold dbt bug) to the correct fix path. - Investigation workflow: Provides a four-step procedure to compare Raw data against source systems and isolate extraction or dbt model bugs. - Use Case: A stakeholder asks to run an UPDATE on a Silver table to remove an employee record. The Skill refuses the request and redirects the fix to the source system (Protheus or RM), letting the pipeline propagate the correction. ## Quick Start Ask the assistant to evaluate a request to fix incorrect employee data in a SIAN Silver table and determine the correct course of action.

Frequently Asked Questions about sian-data-correction-policy

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I fix incorrect data in BigQuery Silver or Gold tables?▼

Never fix data directly in Silver or Gold tables. Correct the data in the source system such as Protheus or RM, and the pipeline will propagate the fix on the next execution. Direct edits are overwritten and create divergence.

What should I do when Raw data differs from the source system?▼

A divergence between Raw and the source system indicates an extraction bug. Fix the bug in the DAG or factory configuration, then reprocess the extraction rather than editing the Raw table directly.

Can I add a CASE WHEN in a dbt model to ignore a bad record?▼

No. Adding CASE WHEN overrides in dbt models to hide records is explicitly refused under this policy. If Silver or Gold diverges from correct Raw data, fix the dbt model logic itself instead of masking individual records.

Why are direct data corrections forbidden in derived data layers?▼

Raw, Silver, and Gold are derived layers rebuilt by pipelines and dbt. Any direct edit is overwritten on the next run, creates divergence from the source system, and makes clean reprocessing impossible.

How do I investigate a data inconsistency across platform layers?▼

Compare the Raw data against the source system first. If Raw diverges, fix the extraction DAG; if Raw is correct but Silver or Gold diverges, fix the dbt model. Never edit data to close an investigation.