A curated collection of Agent Skills for the datavault4dbt dbt package. These skills help AI agents build Data Vault 2 warehouses correctly — staging, hubs, links, satellites, and business-vault entities — across all adapters datavault4dbt supports.
Maintained by Scalefree.
Agent Skills are folders of instructions, examples, and resources that agents discover and load automatically. They are not slash commands: once installed, the agent loads the relevant skill when your prompt matches its use case. Just describe what you need in natural language.
- Building Data Vault models with datavault4dbt: how to lay out a project and generate staging, hubs, links, satellites, and business-vault models using the package's macros and YAML-metadata pattern — with the right hash configuration, naming conventions, and materializations.
These skills are distributed as plain skill folders from this repository — there is no plugin
marketplace or registry listing. Install them by cloning the repo and pointing your agent's skills
directory at skills/datavault4dbt/skills/.
git clone https://github.com/ScalefreeCOM/datavault4dbt-agent-skills.git
cd datavault4dbt-agent-skillsClaude Code auto-discovers skills in two places:
- User scope —
~/.claude/skills/, available in every project. - Project scope —
<your-dbt-project>/.claude/skills/, loaded once you trust the workspace.
Symlinking keeps the skills up to date with git pull:
# User scope (available everywhere)
mkdir -p ~/.claude/skills
ln -s "$PWD"/skills/datavault4dbt/skills/* ~/.claude/skills/# Project scope (only in one dbt project)
mkdir -p /path/to/your-dbt-project/.claude/skills
ln -s "$PWD"/skills/datavault4dbt/skills/* /path/to/your-dbt-project/.claude/skills/Prefer a copy if you want a frozen snapshot instead of a live link:
cp -r skills/datavault4dbt/skills/* ~/.claude/skills/Install a single skill by naming it instead of globbing:
ln -s "$PWD"/skills/datavault4dbt/skills/using-datavault4dbt ~/.claude/skills/Start a session in a dbt project and ask something like "add a customer hub with datavault4dbt" — the
agent should load using-datavault4dbt on its own. Editing a SKILL.md takes effect immediately in the
next session.
Any agent that supports the Agent Skills format works the same way: point
its skills directory at the folders under skills/datavault4dbt/skills/. Check your client's docs for
where that directory lives.
git pull # symlinked installs pick this up automaticallyIf you copied instead of symlinking, re-run the cp after pulling.
| Skill | Status | Description |
|---|---|---|
using-datavault4dbt |
✅ shipped | Build Data Vault models with datavault4dbt — project layout, choosing entities, and the YAML-metadata macro pattern. Entry point that routes to detailed references (staging, hubs/links, satellites, business vault, conventions). |
configuring-datavault4dbt |
✅ shipped | Install the package, copy global variables into dbt_project.yml, hash/naming settings, per-adapter setup. |
testing-a-datavault4dbt-project |
✅ shipped | Data Vault 2 technical tests — hashkey uniqueness/not-null, link→hub referential integrity, satellite key+load-date uniqueness. |
rehashing-datavault4dbt-entities |
✅ shipped | Recalculate hashkeys/hashdiffs after a hash-config change or v1→v2 upgrade, safely and in order. |
troubleshooting-datavault4dbt |
✅ shipped | Diagnose common failures — change-detection, high-water-mark, ghost-record, and YAML-metadata issues. |
building-business-vault-with-datavault4dbt |
🛠 planned (under consideration) | Standalone PIT / snapshot-control skill, if PIT work proves common (currently a using-datavault4dbt reference). |
Most skills assume:
- dbt is installed and a project with
dbt_project.ymlexists. - The datavault4dbt package is (or will be) installed
via
packages.yml. - Basic familiarity with dbt (models, sources, packages) and Data Vault 2 concepts.
These skills work with any AI agent that supports the Agent Skills format (Claude Code, Cursor, and others).
See the Contributing Guide. All skills follow the Agent Skills specification.
See LICENSE for details (Apache-2.0).