AI diagram generator

AI Kubernetes architecture diagram generator.

The topology is in the YAML. It is just spread across forty files and expressed as label matches.

Download for macOS
v0.1.33 · .dmg · Apple Silicon & Intel
Signed & notarized by Apple · opens without a Gatekeeper warning
sha256 698955a0187bc039f4c74f5d05a9f10fbb27376a45788a0a241d1326b73873c7
Download for Windows instead
$curl -fsSL https://lucidtrain.com/install.sh | sh

Sources it reads

Plain manifests, Kustomize overlays and rendered Helm output. Rendering first matters: a chart full of unresolved template values produces a diagram of the template rather than of the deployment, which is accurate about the chart and useless for understanding the environment.

How Lucid Train draws one

Describe what you want, or point it at a repository. The agent emits a semantic graph of nodes and edges with no coordinates, and the ELK layout engine computes positions and orthogonal routing. That separation is the reason the output cannot come back with boxes overlapping or arrows crossing through cards, which is the characteristic failure when a language model is asked to place things on a canvas itself. It also means editing is editing the graph: change a component and the layout re-solves rather than drifting out of alignment.

Selectors become edges

The relationship between a Service and its pods is a label match, and label matches are precisely what humans misread: a selector matching nothing looks identical to one matching correctly. Drawing that relationship makes an orphaned Service obvious immediately, which turns this from a documentation exercise into a debugging one.

Stateful workloads drawn as data

StatefulSets with volume claims land in the data tier rather than sitting alongside stateless Deployments. That single distinction is the one people most need from a cluster diagram, because it answers what can be restarted freely and what cannot.

Local, offline, your models

It runs as a desktop application. Against a local Ollama model the whole thing works offline with no API key and no telemetry, which is the practical difference from every browser-based diagram tool: your architecture never leaves the machine.

Kubernetes resources in the diagram

ResourceDrawn as
Deployment, DaemonSetApplication tier component
StatefulSet with PVCData tier component
Job, CronJobWorker component
ServiceThe edge between selector and target
Ingress, Gateway APIEdge tier entry point
ConfigMap, Secret referencesEdges to dependencies
NamespaceA section, when spanning several
HorizontalPodAutoscalerA note on its target
NetworkPolicyRead, noted, does not constrain drawn edges

Try it

shell
$ Read the Kubernetes manifests here and diagram the workloads, the services connecting them, what holds state, and how traffic enters.
FAQ

Questions

Does it connect to my cluster?

No. It reads YAML from disk, so no kubeconfig and no cluster access, which also means it works against clusters you cannot reach.

Does it work with Helm?

Yes, from rendered output. Render the chart with your values first and the diagram reflects the deployment rather than the template.

Can it diagram the whole cluster including system namespaces?

It can, and the result is dominated by components you did not write. Scoping to your own namespaces produces something a person will read twice.

Will it show the service mesh?

Mesh resources declared as manifests are drawn. Routing decisions made by the control plane at runtime are not in the YAML and therefore not in the diagram.

Related