Neo4j Nvl Skill logo

Neo4j Nvl Skill

Organization
neo4j-contrib
neo4j-nvl-skill

Neo4j Visualization Library (NVL) — framework-agnostic graph rendering for the browser. Covers @neo4j-nvl/base (NVL class, nodes/relationships, Canvas vs WebGL renderer), @neo4j-nvl/interaction-handlers (ZoomInteraction, PanInteraction, DragNodeInteraction, ClickInteraction, HoverInteraction, BoxSelectInteraction, LassoInteraction, KeyboardInteraction), and @neo4j-nvl/react (InteractiveNvlWrapper, BasicNvlWrapper, StaticPictureWrapper). Use when rendering a Neo4j graph in a browser, feeding driver results through nvlResultTransformer, choosing Canvas vs WebGL, wiring node/relationship click/hover/drag handlers, or embedding NVL in React, Vite, or vanilla JS apps. Does NOT handle Cypher query authoring — use neo4j-cypher-skill. Does NOT handle driver lifecycle, sessions, or executeQuery setup — use neo4j-driver-javascript-skill. Does NOT handle GraphVisualization/Needle default embed — use @neo4j-ndl/react.

Overview

Publisherneo4j-contrib
Repositoryneo4j-skills
Skill nameneo4j-nvl-skill
Stars
112
Forks
38
Bundled files
2
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 2 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by neo4j-contrib on GitHub. Read the source before you install it.

Installation

Install the Neo4j Nvl Skill AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/neo4j-contrib/neo4j-skills.git /tmp/neo4j-skills
mkdir -p .claude/skills
cp -r /tmp/neo4j-skills/neo4j-nvl-skill .claude/skills/neo4j-nvl-skill
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Neo4j Nvl Skill in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Neo4j Nvl Skill on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Neo4j Nvl Skill is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

When to Use

  • Rendering a Neo4j graph in a browser (vanilla JS, React, Vite) with custom interactions, rendering, or data shapes
  • Visualizing driver.executeQuery results as an interactive graph
  • Wiring zoom, pan, drag, click, hover, lasso, or box-select interactions
  • Embedding NVL inside an existing app and synchronizing graph state

When NOT to Use

  • Pre-styled embedded graph view with default behavior, no custom interactionsGraphVisualization from @neo4j-ndl/react (Neo4j Needle / NDL design system) — wraps NVL with default Neo4j styling. See Use NVL or the Needle Component? below.
  • Python / Jupyter notebook graph visualizationneo4j/python-graph-visualization (the Python port of NVL)
  • Writing/optimizing Cypherneo4j-cypher-skill
  • Driver setup / executeQuery / sessionsneo4j-driver-javascript-skill
  • Server-side data fetching with no renderingneo4j-driver-javascript-skill
  • GDS algorithm executionneo4j-gds-skill or neo4j-aura-graph-analytics-skill
  • GraphQL APIneo4j-graphql-skill

Use NVL or the Needle Component?

NeedUse
Embed a graph view with default Neo4j styling, no custom interactions or renderingGraphVisualization from @neo4j-ndl/react (Neo4j Needle / NDL design system) — wraps NVL and accepts records shaped { id, labels, properties: { key: { stringified, type } } } (NeoNode)
Custom interactions, custom rendering, non-standard data shapes, or framework-agnostic embeddingThis skill — use NVL directly

If the answer is the first row, install and use the Needle component instead of NVL — do not duplicate styling work.


Install

bash
npm install @neo4j-nvl/base                  # core (required)
npm install @neo4j-nvl/interaction-handlers  # standard interactions (optional, vanilla JS)
npm install @neo4j-nvl/react                 # React wrappers (optional)

Peer requirements: React 19 for @neo4j-nvl/react. The published peerDependency range still permits React 18, but mixing major versions is not recommended — target 19. @neo4j-nvl/layout-workers is a transitive dependency — never install directly. neo4j-driver is a peer of @neo4j-nvl/base only when using nvlResultTransformer.

Starter templates: https://github.com/neo4j-devtools/nvl-boilerplates — official per-framework scaffolds; prefer these over hand-rolled setups.

License: NVL ships under the Neo4j Visualization Library License — for use with Neo4j products only. Cannot be used against other graph backends.


Pick the Right Paradigm

NeedUse
React app, default interactions<InteractiveNvlWrapper> from @neo4j-nvl/react
React app, custom interaction wiring<BasicNvlWrapper> + own handlers via ref
Vanilla JS, standard interactionsNVL + @neo4j-nvl/interaction-handlers
Vanilla JS, fully custom event logicNVL + container.addEventListener + nvl.getHits()
Static PNG/SVG image export<StaticPictureWrapper> or nvl.saveToFile() / nvl.saveToSvg()

Pick the Right Renderer

RendererMax nodesDetailUse case
'canvas' (default)~1,000Full captions, icons, arrows, pixel-perfect hit-testingDetail investigation, small graphs
'webgl'100,000+Reduced label fidelity (bound by GPU max texture size)Large-scale pattern exploration
javascript
const nvl = new NVL(container, nodes, rels, { renderer: 'webgl' })
nvl.setRenderer('canvas')   // swap at runtime

Container Setup

The container must have an explicit width AND height. Missing height → container collapses to 0 → graph invisible. Most-reported NVL bug.

html
<!-- ❌ height defaults to 0; graph invisible -->
<div id="viz"></div>

<!-- ✅ explicit dimensions -->
<div id="viz" style="width: 100%; height: 600px;"></div>

Vanilla — Base Library

javascript
import { NVL } from '@neo4j-nvl/base'

const container = document.getElementById('viz')
const nodes = [{ id: '1' }, { id: '2' }]
const relationships = [{ id: '12', from: '1', to: '2', type: 'KNOWS' }]

const nvl = new NVL(container, nodes, relationships)

With options + callbacks:

javascript
import { NVL } from '@neo4j-nvl/base'

const options = {
  initialZoom: 1.0,
  minZoom: 0.1,
  maxZoom: 8,
  layout: 'forceDirected',
  renderer: 'canvas',
  styling: { defaultNodeColor: '#0e86d4', defaultRelationshipColor: '#888' }
}
const callbacks = {
  onInitialization: () => console.log('NVL ready'),
  onLayoutDone: () => nvl.fit([]),
  onError: (err) => console.error('NVL error', err)
}

const nvl = new NVL(container, nodes, relationships, options, callbacks)

// On teardown — always:
nvl.destroy()

NVL constructor signature: new NVL(frame, nvlNodes?, nvlRels?, options?, callbacks?). All but frame are optional and default to empty.


Vanilla — Interaction Handlers

Compose handlers onto an existing NVL instance. Each handler registers callbacks via .updateCallback(name, fn) and must be torn down with .destroy().

javascript
import { NVL } from '@neo4j-nvl/base'
import {
  ZoomInteraction, PanInteraction, DragNodeInteraction,
  ClickInteraction, HoverInteraction, BoxSelectInteraction,
  LassoInteraction, KeyboardInteraction
} from '@neo4j-nvl/interaction-handlers'

const nvl = new NVL(container, nodes, relationships)

const zoom  = new ZoomInteraction(nvl)
const pan   = new PanInteraction(nvl)
const drag  = new DragNodeInteraction(nvl)
const click = new ClickInteraction(nvl, { selectOnClick: true })
const hover = new HoverInteraction(nvl, { drawShadowOnHover: true })

click.updateCallback('onNodeClick',         (node, hits, evt) => console.log('node',  node.id))
click.updateCallback('onRelationshipClick', (rel,  hits, evt) => console.log('rel',   rel.id))
click.updateCallback('onCanvasClick',       (evt)             => console.log('canvas'))
hover.updateCallback('onHover',             (el, hits, evt)   => el && console.log('over', el.id))
drag.updateCallback('onDragEnd',            (nodes, evt)      => savePositions(nodes))
zoom.updateCallback('onZoom',               (level)           => console.log('zoom', level))

// Teardown — destroy all handlers, then the NVL instance
function teardown() {
  for (const h of [zoom, pan, drag, click, hover]) h.destroy()
  nvl.destroy()
}

Disable an event without removing the handler: click.removeCallback('onCanvasClick'). Passing true instead of a function enables the event with a no-op (useful for default selection behavior).


React — InteractiveNvlWrapper

Pre-wires every interaction handler. Toggle events with mouseEventCallbacks (function = on + callback; true = on, no-op; false/omit = off).

tsx
import { InteractiveNvlWrapper } from '@neo4j-nvl/react'
import type { MouseEventCallbacks, NvlOptions } from '@neo4j-nvl/react'
import { useRef } from 'react'
import type { NVL } from '@neo4j-nvl/base'

export function GraphView({ nodes, rels }) {
  const nvlRef = useRef<NVL>(null)

  const nvlOptions: NvlOptions = { initialZoom: 1, renderer: 'canvas' }

  const mouseEventCallbacks: MouseEventCallbacks = {
    onNodeClick:         (node, hits, evt) => console.log('node',  node.id),
    onRelationshipClick: (rel,  hits, evt) => console.log('rel',   rel.id),
    onCanvasClick:       (evt)             => console.log('canvas'),
    onHover:             (el, hits, evt)   => el && console.log('hover', el.id),
    onDragEnd:           (nodes, evt)      => persist(nodes),
    onZoom: true,                                       // enable, no callback
    onPan:  true
  }

  return (
    <div style={{ width: '100%', height: 600 }}>
      <InteractiveNvlWrapper
        ref={nvlRef}
        nodes={nodes}
        rels={rels}
        nvlOptions={nvlOptions}
        interactionOptions={{ selectOnClick: true, drawShadowOnHover: true }}
        mouseEventCallbacks={mouseEventCallbacks}
        onInitializationError={(err) => console.error('NVL init', err)}
      />
    </div>
  )
}

ref resolves to the underlying NVL instance — call any method on it: nvlRef.current?.fit([]), nvlRef.current?.setRenderer('webgl'), nvlRef.current?.saveToFile().


React — BasicNvlWrapper + Ref

No interactions wired. The ref exposes every NVL method via IncludeMethods<NVL> — use when building custom interaction logic in React.

tsx
import { BasicNvlWrapper } from '@neo4j-nvl/react'
import type { NVL } from '@neo4j-nvl/base'
import { useRef } from 'react'

export function MiniGraph({ nodes, rels }) {
  const nvlRef = useRef<NVL>(null)

  return (
    <div style={{ width: '100%', height: 400 }}>
      <BasicNvlWrapper
        ref={nvlRef}
        nodes={nodes}
        rels={rels}
        nvlOptions={{ initialZoom: 2 }}
        nvlCallbacks={{ onLayoutDone: () => nvlRef.current?.fit([]) }}
      />
      <button onClick={() => nvlRef.current?.fit(['1', '2'])}>Zoom to 1,2</button>
    </div>
  )
}

Wiring a Neo4j Driver Result

@neo4j-nvl/base exports a ResultTransformer for the JS driver that deduplicates nodes/relationships across any record shape.

javascript
import neo4j from 'neo4j-driver'
import { NVL, nvlResultTransformer } from '@neo4j-nvl/base'

const driver = neo4j.driver(process.env.NEO4J_URI,
  neo4j.auth.basic(process.env.NEO4J_USERNAME, process.env.NEO4J_PASSWORD))

const { nodes, relationships } = await driver.executeQuery(
  'MATCH (a)-[r]-(b) RETURN a, r, b LIMIT 25',
  {},
  { database: 'neo4j', resultTransformer: nvlResultTransformer }
)

const nvl = new NVL(document.getElementById('viz'), nodes, relationships)
javascript
// ❌ raw EagerResult — records are not Node/Relationship objects
const result = await driver.executeQuery('MATCH (a)-[r]-(b) RETURN a, r, b')
new NVL(container, result.records, [])   // breaks

// ✅ use the transformer
const { nodes, relationships } = await driver.executeQuery(
  'MATCH (a)-[r]-(b) RETURN a, r, b',
  {},
  { database: 'neo4j', resultTransformer: nvlResultTransformer }
)
new NVL(container, nodes, relationships)

For driver lifecycle, session management, Integer handling, and TypeScript types → neo4j-driver-javascript-skill.


Updating the Graph

MethodBehavior
addAndUpdateElementsInGraph(nodes, rels)Insert new; update existing by id (only specified fields)
updateElementsInGraph(nodes, rels)Update existing only; ignores unknown ids
addElementsToGraph(nodes, rels)Insert only; throws on existing id
removeNodesWithIds(ids)Remove nodes; adjacent relationships auto-removed
removeRelationshipsWithIds(ids)Remove relationships
setNodePositions(nodes, updateLayout?)Override positions; optionally re-run layout
restart(options?, retainPositions?)Restart with new options; positions optional

Diff updates use PartialNode / PartialRelationship — only id is required:

javascript
nvl.updateElementsInGraph(
  [{ id: '1', color: '#f00', selected: true }],   // PartialNode
  [{ id: '12', width: 4 }]                         // PartialRelationship
)

Hit Testing (Manual)

Use when NOT using the interaction-handlers package. getHits() resolves which node/relationship is under a pointer event.

javascript
const nvl = new NVL(container, nodes, rels)

container.addEventListener('click', (evt) => {
  const { nvlTargets } = nvl.getHits(evt, ['node', 'relationship'], { hitNodeMarginWidth: 4 })
  const hitNode = nvlTargets.nodes[0]
  const hitRel  = nvlTargets.relationships[0]
  if (hitNode) console.log('hit node', hitNode.data.id)
  else if (hitRel) console.log('hit rel', hitRel.data.id)
  else console.log('hit canvas')
})

HitTargetNode / HitTargetRelationship carry data, pointerCoordinates, distance, insideNode (nodes only). See references/api-surface.md.


Common Mistakes

MistakeFix
Container with no height → invisible graphSet explicit width and height on the container
Pass driver.executeQuery result directlyUse nvlResultTransformer and consume { nodes, relationships }
WebGL for small label-rich graphsUse 'canvas'; labels are fully supported
Canvas for 10k+ nodesSwitch to 'webgl' via renderer option or setRenderer
New NVL per React renderUse <InteractiveNvlWrapper> / <BasicNvlWrapper> or wrap in useEffect + destroy()
Forgetting nvl.destroy() on teardownCall destroy() on unmount; React wrappers handle this automatically
Vanilla handlers not torn downCall .destroy() on every interaction before nvl.destroy()
Worker construction blocked (strict CSP / sandboxed runtime / older bundler)nvlOptions: { disableWebWorkers: true } (NVL has a non-worker fallback)
Telemetry enabled in regulated envnvlOptions: { disableTelemetry: true }
Layout never settlesPin anchor nodes with pinNode(id); tune layoutTimeLimit
selectOnClick fires doubleToggle once at mount; don't flip interactionOptions per render
Hit test misses near node edgePass { hitNodeMarginWidth: N } to getHits
Captions missing on WebGLGPU max texture size exceeded; fall back to Canvas or shrink captions

References

Load on demand:

  • references/api-surface.md — complete NVL method table; Node, Relationship, NvlOptions, LayoutOptions, ExternalCallbacks, HitTargets, NvlMouseEvent, StyledCaption, Point; every interaction-handler class + its options + its callback signatures; React <InteractiveNvlWrapper> / <BasicNvlWrapper> / <StaticPictureWrapper> props; MouseEventCallbacks and KeyboardEventCallbacks shapes; named exports inventory; nvlResultTransformer signature
  • references/troubleshooting.md — zero-height container, build-tool-agnostic disableWebWorkers fallback, Canvas/WebGL trade-offs + WebGL2 note, WebGL texture-size cap, onWebGLContextLost recovery, telemetry opt-out, memory leaks, stuck layouts, double selection, hit-margin tuning, license restriction

Canonical web documentation (use WebFetch when references above are insufficient):


Checklist

  • Container has explicit width AND height CSS
  • Correct paradigm chosen from the decision table (vanilla / handlers / React)
  • Renderer matches expected node count (Canvas ≲1k / WebGL 100k+)
  • Driver executeQuery results piped through nvlResultTransformer
  • database specified on every executeQuery call (delegate to neo4j-driver-javascript-skill)
  • All interaction handlers .destroy()-ed before nvl.destroy() on teardown
  • nvl.destroy() called on React unmount (manual instances only — wrappers handle it)
  • disableTelemetry: true set when in regulated / offline environments
  • disableWebWorkers: true set when bundler / CSP blocks worker construction
  • Graph updates use addAndUpdateElementsInGraph / updateElementsInGraph — not restart
  • License compatible: target is a Neo4j product

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Neo4j Nvl Skill AI skill do?

Neo4j Visualization Library (NVL) — framework-agnostic graph rendering for the browser. Covers @neo4j-nvl/base (NVL class, nodes/relationships, Canvas vs WebGL renderer), @neo4j-nvl/interaction-handlers (ZoomInteraction, PanInteraction, DragNodeInteraction, ClickInteraction, HoverInteraction, BoxSelectInteraction, LassoInteraction, KeyboardInteraction), and @neo4j-nvl/react (InteractiveNvlWrapper, BasicNvlWrapper, StaticPictureWrapper). Use when rendering a Neo4j graph in a browser, feeding driver results through nvlResultTransformer, choosing Canvas vs WebGL, wiring node/relationship click...

Why use Neo4j Nvl Skill on TypingMind?

Because you install it once and use it with any model. Neo4j Nvl Skill is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Neo4j Nvl Skill in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/neo4j-contrib/neo4j-skills/tree/main/neo4j-nvl-skill. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Neo4j Nvl Skill?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Neo4j Nvl Skill?

As many as you like. As long as a model supports skills, you can use Neo4j Nvl Skill with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Neo4j Nvl Skill AI skill free?

Yes. It is published on GitHub by neo4j-contrib under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇