dbt-pipeline-toolkit

Business analyst specialist for the dbt-pipeline-toolkit. Reads every data profile in `1 - Documentation/data-profiles/` and drafts the source-aware discovery question set for the orchestrator to ask. Operates in two modes: `prepare` (return the structured question set as a JSON envelope) and `write` (consume the user's answers, supplied by the orchestrator, and write Section 1 of `pipeline-design.md` plus any headerless-profile rewrites). NEVER calls AskUserQuestion — that tool is main-thread-only and unavailable to subagents; the orchestrator owns all user interaction. Invoked by `dbt-pipeline-orchestrator` at Stage 2.

Autonomous data discovery agent that profiles source tables, summarizes schemas, and maps relationships. Runs data-profiler scripts, reads existing profiles, and returns concise structured summaries. Use when you need to UNDERSTAND source data before building models -- not for requirements gathering (use business-analyst) or ad-hoc queries (use sql-server-reader). Runs in background, no user interaction.

Initialize new data engineering projects with the complete folder structure, dbt configuration, Python virtual environment, and CLAUDE.md for agentic development. Use proactively when starting a new analytics/BI project, creating a data pipeline repository from scratch, or scaffolding a dbt + SQL Server + Power BI project. Sets up everything needed for personal agents to start implementing.

Build dimension tables (dim_*) with attributes and hierarchies. Handle surrogate keys, natural keys, and SCD patterns (Type 1 and Type 2). Organize attributes logically and create role-playing dimensions. MUST BE USED when creating dimension tables for star schema models.

Build fact tables (fct_*) with measures at specific grain. Implement incremental materialization strategies optimized for SQL Server (delete+insert, merge, append). Generate surrogate keys and create foreign key relationships to dimensions. Handle large transaction datasets efficiently. MUST BE USED when creating fact tables for star schema.

Automatically profile SQL Server tables and CSV files with intelligent analysis. Detects primary key candidates, infers data types from CSV data, calculates column statistics (nulls, cardinality, data types), identifies data quality issues, and recommends appropriate dbt tests. Use when exploring source data, creating staging models, or validating data quality before transformation. Generates comprehensive profiling reports with test recommendations.

Generate and serve dbt documentation including model lineage, column descriptions, and test results. Use when creating dbt documentation, viewing model relationships, exploring data lineage, or sharing documentation with stakeholders. Supports generating static sites, serving locally, and exporting documentation artifacts.

Initialize a new data engineering project with the complete folder structure, dbt configuration, Python virtual environment, and CLAUDE.md for agentic development. Use when starting a new analytics/BI project, creating a data pipeline repository, or scaffolding a dbt + Power BI project from scratch. Sets up everything needed for personal agents and skills to start implementing pipelines.

Execute dbt commands (run, test, compile, docs, build, snapshot, seed, source freshness, deps, debug, list, clean) for analytics engineering workflows. Use when running dbt models, testing data quality, generating documentation, managing dbt projects, running snapshots, or implementing Slim CI patterns. Supports advanced model selection, full refresh, state comparison, and parallel execution.

Analyze dbt test coverage across all models and identify gaps. Calculate coverage percentages, find untested models, identify missing primary key and foreign key tests. Use when validating test coverage, ensuring data quality standards, or preparing for production deployment. Reports coverage by layer (staging, marts) and model type.

4 hooks across 3 events