SkillFoxx

Dex plugin

The Dex plugin: an agent explores a database, builds dbt models and a semantic layer and tracks schema drift under cost control

  • 119 ★ on the repository
  • Last commit
  • Apache-2.0
Plugin

Extended access

Why this level

  • Dex connects to databases and warehouses, and billed connectors spend money on your account
  • Commands apply changes to a dbt project and build models on a dev target
All reasons and checks

exmergo/dexAuthor:Exmergo

Install

In your terminal, with SkillFoxx CLI

npx skillfoxx add plugins/exmergo-dex

Detects the agents on your machine, shows the access level and pins the version.

Other ways to install

Access and checks

How we review

Extended accessThe tool runs code or makes network calls and only reads external data.

  • Dex connects to databases and warehouses, and billed connectors spend money on your account
  • Commands apply changes to a dbt project and build models on a dev target

This is third-party code. Review the repository files before installing.

What it does

Dex gives an agent three skills and the `exmergo-dex-core` engine that together run analytics through an explore, transform, maintain loop. In Claude Code they are `/dex:explore`, `/dex:transform` and `/dex:maintain`, and Codex gets its own plugin that needs no MCP server. Exploration writes nothing to the data: Dex finds the tables that matter, profiles columns, flags personal data, verifies joins, draws a Mermaid relationship diagram and answers ad-hoc questions through a query firewall that keeps values such as email addresses out of the model's context. Transform builds dbt models from staging to marts with tests and docs, plus metrics on a semantic layer. A plan is reviewed first, then applied and verified on a dev target. Maintain compares the database, the project and the layer with a snapshot, shows which models a change affects and proposes edits. Connectors: Snowflake, BigQuery, Databricks, Redshift, Postgres, DuckDB, ClickHouse and Trino. Before any billed scan an estimate is shown, execution follows your agreement and a spend cap is enforced server-side. You can try it with no database of your own through `dex demo`.

Who it is for. Analytics engineers and data teams who run a dbt project and want an agent to handle database exploration, models and change tracking.

Good fit when

  • A new database needs a map of which tables matter, how they join and where personal data sits
  • A dbt model must be built from staging to a mart with tests, docs and a declared grain
  • A source schema changed and you need to see which models and metrics are affected and the proposed edits
  • Before scanning a big warehouse you want a cost estimate and a hard spend cap

Not a fit when

  • You use an agent with no local shell: Dex runs through the command line and `uv`
  • You need data in Excel or Python tables for a one-off job: the engine targets an analytics engineering project
  • You need a connector for MotherDuck, Starburst Galaxy or Microsoft Fabric: they are planned and not ready yet

Example request

Run dex demo and dex explore map, then show which tables were found and which columns are flagged as personal data.

Limitations

You need `uv` 0.13.4 or newer installed and an agent with a local shell, and no agent installs `uv` for you. Source data stays read-only and builds run on a dev target. Dex does not let values from personal-data columns into context but allows aggregates such as `count(distinct email)`. Billed connectors spend money on your account, so a scan needs confirmation first and a spend cap is held server-side. MotherDuck, Starburst Galaxy and Microsoft Fabric connectors do not exist yet. The core is released under Apache-2.0, and a paid Dex Enterprise exists too.

How to disable. Remove the dex plugin through /plugin or the skills from the agent's skills directory. Remove the exmergo-dex-core package through uv or pip, and the .dex directory with cache and snapshots from the project by hand.

FAQ

How do I install Dex?

For any agent run `npx skills add exmergo/dex`. In Claude Code: `/plugin marketplace add exmergo/exmergo-agent-plugins`, then `/plugin install dex@exmergo`. For Codex: `npx skills add exmergo/dex --agent codex`.

How can I try it without my own database?

Install `pip install "exmergo-dex-core[duckdb]"`, run `dex demo`, then `dex explore map`. The demo creates a DuckDB database with 7 tables and 29,512 rows plus `.dex/config.yml`.

Does the agent see personal data?

No. The query firewall refuses to put values from columns flagged as personal data into context, such as `select email from customers`. An aggregate like `count(distinct email)` runs because it is a statistic.

How does Dex limit cost?

Every scan is estimated, confirmed and capped server-side, and settled spend goes to a local ledger. Self-hosted databases bill no dollars but are still gated, because an unbounded scan on a production Postgres is costly.

Which semantic layers are supported?

dbt semantic models with MetricFlow, plus Apache Ossie, SLayer and Cube files. The choice is set in `.dex/config.yml`, and Ossie, SLayer and Cube need no dbt project.

Discussion

No questions or reviews yet.

Plugins

DataMind gives Codex a shared data layer: knowledge base, SQL, graph and memory, with a receipt for every write.

PluginExtended access455

Ragstar answers questions about a dbt project in chat and Slack, and its MCP server gives Claude search over models and their SQL

MCP serverExtended access180
Official

A Google plugin for Claude Code, Codex and Antigravity: Looker data queries, dashboards and LookML in plain language.

PluginExtended access58

A Claude Code agent for protein design: target research, antibody and nanobody generation, screening and candidate ranking.

PluginExtended access120

A Rust terminal database client with MCP and a JSON CLI: agents read schemas and run queries within profile permissions.

CLIExtended access67

Skills for Claude Code on dbt and Snowflake: models, tests, debugging, migrations, query tuning and handoff to altimate-code

PluginExtended access136

README in short

The README opens with three capabilities: explore, transform and maintain. Installation for agents, a demo on a deliberately broken database, the loop description, tables of connectors and semantic layers, using the engine from Python, two benchmark results, lifecycle management and Dex Enterprise notes follow.

README badge

Original SKILL.md

---
name: explore
description: 'Use this whenever you need to know what is actually in a database, warehouse, or DuckDB file before you trust it: inventory, column profiles, PII detection, grain and data-quality problems, verified joins, ER diagrams, ad-hoc SQL probes, segmentation, and the semantic layer a repo decl
compatibility: 'Needs a local shell and uv. On first use, uv builds the exmergo-dex-core engine from the source bundled with the plugin, or installs it from PyPI at the version scripts/run.py pins when no source is bundled. Runs in Claude Code or another agent that loads Agent Skills and has a shell
---

# Explore

Make sense of a warehouse or a local DuckDB database the way an analytics
engineer does: rank what matters, drill selectively, and persist a draft map in

Report a problem