dbt-workflows

Routes dbt intent through governed Seshat shadow workflows with evidence and approval gates.

2|Updated Jun 22, 2026
One-click install
npx skills add https://github.com/Kemetra/Seshat-BI --skill dbt-workflows-kemetra
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: dbt-workflows
Source: https://github.com/Kemetra/Seshat-BI/tree/main/distribution/bundle-templates/shared/skills/dbt-workflows
Command: npx skills add https://github.com/Kemetra/Seshat-BI --skill dbt-workflows-kemetra

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Running dbt against a governed analytics project risks bypassing readiness gates, breaking mapping approvals, and producing unverifiable results. This Skill routes every dbt intent to the correct owner and enforces Seshat's fixed selector, target, plan, and evidence workflow. ## Core Features & Use Cases - Intent Routing: Directs dbt tasks to Seshat's governed shadow workflow, the official dbt Labs agent skills, or the official dbt MCP based on the user's intent. - Governed Execution Pipeline: Runs the fixed sequence of doctor, scaffold, validate, plan, build, and inspect-run commands with immutable plan digests and drift detection. - Contract Enforcement: Applies the meta.seshat model contract so every model cites a committed source map revision and every output column carries a citation or governed derivation. - Use Case: A data engineer needs to build a governed gold fact table from an approved source map. The Skill scaffolds staging and gold models, validates the Mapping Ready gate, produces a reviewable plan, and executes the build only after the digest is accepted. ## Quick Start Ask the assistant to run seshat dbt doctor and then scaffold, validate, plan, and build the governed dbt models for a specific table.

Frequently Asked Questions about dbt-workflows

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

FAQPage Schema
How do I run a governed dbt build with Seshat?▼

Run seshat dbt doctor, scaffold, validate, and plan for the table, review the returned plan digest, then execute seshat dbt build with --accept-plan set to that digest. Drift between plan and build stops execution and requires a new plan.

When should dbt tasks go to the official dbt Labs skills instead of Seshat?▼

Generic dbt model authoring, test design, syntax, docs, and troubleshooting route to the official dbt Labs agent skills after proving they are discoverable via seshat integrations setup. Only governed execution against the Seshat project uses seshat dbt commands.

What is the meta.seshat model contract in dbt scaffold output?▼

Each generated _models.yml row carries top-level meta.seshat fields including table_id, source_map, source_map_revision, grain, business_key, and authority. Every output column must cite source_columns or one of four governed derivation values, or validate reports a defect.

Why does seshat dbt validate report DBT_MODEL_CITATION_STALE?▼

The model's source_map_revision no longer matches the git blob sha of the committed source map. Re-commit the map and update each model's source_map_revision by hand, or delete the generated _models.yml and re-scaffold.

Can I run raw dbt commands against the governed Seshat project?▼

No. Raw dbt commands bypass the fixed selector, target, profile, lock, redaction, and artifact checks. All governed compile, build, test, and parity work must go through the seshat dbt helpers.

What do the seshat dbt exit codes mean?▼

Exit 0 means the command completed with derived evidence, 1 is a handled model or parity failure, 2 is a missing prerequisite, 3 is a governance refusal or plan drift, and 4 is an artifact integrity failure.