Getting started

Electros LaunchPad is the local screen for VMs, AI models, and marketplace services. You do not SSH into a host to start: you pick an action here, and LaunchPad talks to the local daemon (elpd) for you.

The sidebar is the map. The orange row is the page you are on. Badges show live counts (instances, running models, credentials, deployments). The footer is the signed-in account and daemon health.

What the shell is for

LaunchPad is a local control plane:

• The sidebar never creates a resource. It only moves you.
• The page on the right is where you read inventory or change it.
• The footer is the first thing to check when a list is empty or a button does nothing. All systems healthy means the daemons LaunchPad depends on are responding. If that line turns red, fix health before debugging the form.

The sidebar

Area	Pages	What you do there
Top	Home, Compositions	Fleet glance; named groups of workloads
Compute	Images, Instances, Cloud-init	Browse OS images, manage VMs, edit user-data
AI	Models, Downloaded, Running models, Runner, Credentials	Find models, load them, pick accelerators, manage keys
Services	Catalogue, Deployments	Deploy ready-made tools into their own VMs
Tools	Installers, Storage, Migration Hosts, Routing, Help	Binaries, disk usage, peer hosts, port forwards

Home dashboard

Home answers three questions: what can I create, what is running, and how loaded is this machine?

Primary actions

+ New VM, Run model, and Deploy service jump into the three main create paths without hunting the sidebar.

The workloads pill summarises how many VMs, LLMs, and services are running right now.

Running now

Filter with All / VMs / LLMs / Services. Each card shows state, resource use, and identity (IP for VMs, backend for models). Toggle Group by composition when workloads belong to a named group.

System overview (right rail)

Host identity, CPU / memory / disk gauges, available capacity, and the Core allocation grid (claimed VM cores vs free). Use this before launching another heavy VM or model.

The bottom of Home also offers three entry panels — Create, AI, and Services — that deep-link into Images / Instances, Downloaded / Running models, and Catalogue / Deployments.

A sensible first path

1. Confirm the sidebar footer shows All systems healthy.
2. Open Installers and verify llmfit, llama.cpp, and MLX show Ready (see System tools).
3. Launch a small VM from Compute → Images (Alpine or Ubuntu). See Compute.
4. Open the instance Shell and run a command.
5. Download a small chat model, Load it, then try a prompt. See AI models.
6. Create an sk-elp- key under Credentials if another client will call the gateway. See Credentials & runner.

Next

• Compositions — wire VMs, LLMs, and services together
• Tutorial — the same path as a timed walkthrough