Generate Validation Notebook

What it does

This skill generates a YAML-formatted SQL notebook of validation queries for changed dbt models. It analyzes changed .sql files (PR or local), infers schemas using a resolver script, classifies models as new or modified, and emits parameterized queries for row counts, sample previews, uniqueness, null rates, time-series continuity, and before/after comparisons. The output is an import URL for MC Bridge notebooks.

When to use it

Use when reviewing dbt pull requests or local branches to quickly generate a reproducible set of validation queries for data engineers and reviewers. Triggered by a GitHub PR URL or a local dbt repo path.

What's included

• Scripts: includes helper scripts (resolve_dbt_schema.py and generate_notebook_url.py) under scripts/ for schema resolution and URL generation (has_scripts=true)
• References: no separate references directory (has_references=false)
• Instructions: detailed phases covering changed-file discovery, schema resolution, diff analysis, query generation patterns, notebook YAML structure, and URL generation.

Compatible agents

Works with agents that can run gh (GitHub CLI) and python3 helper scripts; useful with Claude Code, Sonnet-enabled flows, and other developer-focused agent runtimes.