dbt-nova

Designs, debugs, and operationalizes dbt-nova eval suites for manifest metadata quality and agent tool-use quality. Use when creating bridge evals, provider-backed agent evals, CI eval gates, regression suites for Nova metadata, or when interpreting dbt-nova eval artifacts and failures.

Adversarially reviews high-stakes dbt-nova analytics answer drafts for evidence quality, semantic-layer bypass, unsafe raw joins, stale or unknown freshness caveats, and provenance-backed fixes. Use before leadership-bound, launch-readiness, customer-facing, or recurring production answers.

Answers business questions with Nova using first-principles decomposition, canonical KPI discovery, explicit entity selection, validated filters, bounded execution, and evidence-first reporting. Use when resolving metrics, comparing performance, validating measures, or running recurring analytical workflows through Nova MCP or the dbt-nova CLI.

Runs deterministic governance audits with Nova using frozen scope, repeatable scoring, explicit blocker extraction, and remediation planning. Use when enforcing metadata standards, testing and documentation gates, ownership and compliance requirements, and rerunnable audit evidence through MCP or the dbt-nova CLI.

Builds and reviews high-signal `meta.nova` and curated domain references for dbt resources. Use when adding Nova metadata, choosing canonical datasets or indicators, defining model-bound measures or metric templates, annotating columns, de-ranking helper models, authoring domain reference docs, or improving Nova MCP search and metadata scores.

Improves dbt project structure with Nova using first-principles architecture work: define the repeated business concept, scope candidate clusters, compare overlap, grain, semantics, lineage, and blast radius, choose canonical execution surfaces, and produce migration-safe refactor plans through Nova MCP or the dbt-nova CLI.

Finds and plans structural cleanup in dbt projects with Nova using first-principles triage: freeze scope, detect overlap clusters, duplicate indicators, naming drift, and canonical conflicts, separate accidental duplication from legitimate specialization, quantify impact, and produce prioritized cleanup queues through Nova MCP or the dbt-nova CLI.

Designs dashboard-ready analytical products with Nova using first-principles BI engineering: start from the decision, audience, and consumer artifact; resolve canonical indicators; choose execution entities and grain; validate filters, comparisons, and bounded sample outputs; then produce dashboard specs, dataset contracts, metric cards, and viz QA handoffs through MCP or the dbt-nova CLI.

Builds and modifies dbt models with Nova using first-principles engineering: change classification, reuse before rebuild, explicit grain and contract checks, blast-radius analysis, proportionate quality gates, manifest readiness, and ship-safe summaries. Use when changing model SQL, columns, grain, metadata, tests, or semantic contracts through Nova MCP or the dbt-nova CLI.

Investigates KPI discrepancies with Nova using first-principles debugging: restate the discrepancy, resolve the canonical KPI, validate time/filter/comparison contracts, reproduce one bounded canonical result, then isolate root causes across definition, grain, filters, lineage, freshness, and alternate sources through Nova MCP or the dbt-nova CLI.