# Kedroviz Specialist

> Use when visualizing Kedro pipeline DAGs, exploring node dependencies, or generating pipeline documentation. Invoke when the user asks about pipeline structure or data flow.

- Skill: `deagentic/kedroviz-specialist` (Agent Skill)
- Install (CLI): `npx skillmds@latest add deagentic/kedroviz-specialist`
- Raw SKILL.md: https://api.skillmd.com/api/skills/deagentic/kedroviz-specialist/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: DevOps & Infra
- Author: deagentic (https://skillmd.com/u/deagentic)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/deagentic/kedroviz-specialist

---


# KedroViz Specialist

## Identity

This skill handles Kedro pipeline visualization using KedroViz. It helps teams understand pipeline structure, validate node dependency ordering, and generate DAG documentation for review and onboarding.

## Scope

- **DAG visualization**: Running `kedro viz` to launch the interactive pipeline explorer.
- **Mermaid export**: Generating Mermaid flowchart representations of pipelines for embedding in documentation.
- **Dependency validation**: Confirming that node input/output names are correctly wired (no dangling inputs, no orphaned outputs).
- **Pipeline documentation**: Generating `docs/pipelines/<pipeline_name>.md` with node descriptions, input/output datasets, and parameter references.

## Usage

```bash
# Launch interactive KedroViz
kedro viz

# Export pipeline as JSON (for programmatic inspection)
kedro pipeline describe --pipeline <pipeline_name>
```

## Mermaid DAG Format

For documentation commits, export pipelines in Mermaid format:

```mermaid
flowchart LR
  raw_ventas[(raw_ventas)] --> node_ingest_ventas
  node_ingest_ventas --> clean_ventas[(clean_ventas)]
  clean_ventas --> node_transform_ventas
  node_transform_ventas --> feature_ventas[(feature_ventas)]
```

## Status

This skill is a stub. Full implementation will include automated Mermaid export, DAG diff tooling for PR reviews, and integration with the `learning-protocol` to persist pipeline structure discoveries.

## Related Skills

- `kedro-builder` — produces the pipelines this skill visualizes
- `deacero-domain` — provides business context for node and dataset naming