Skip to content

WaveGridInfrastructure inventory and cost intelligence

The console in the cloud — resources, rates, alerts. The agent on your side collects data; provider credentials never leave your network.

WaveGrid is a console for teams whose infrastructure is spread across providers: Proxmox, clouds, DNS, monitoring. One workspace shows what you have, what it costs, and what broke or is getting expensive.

Two layers:

  1. The agent runs on your side. Plugins query providers; Proxmox, AWS, and other credentials never leave your network. The console receives normalized snapshots.
  2. The console normalizes resources, tracks rates, evaluates alert rules, and shows the result to the team by role.

Who it is for ​

Role on the teamQuestionSection
DevOpsWhat do we have, and where?Inventory
CTO / financeWhat is expensive, and when do we pay?Costs
OwnerWhat does infrastructure cost by project?Costs + overview

A workspace has three access roles: owner (everything, including settings and alert rules), devops (writes inventory, read-only costs), manager (writes costs, read-only inventory). People join by invite. The agent uses a workspace API token, not a person’s password. Details: Team.

Capabilities ​

Inventory ​

A single catalog: servers, VMs, domains (grouped by apex), storage, and manual records. Filters, tags, one project per resource, bulk edit, CSV. Cross-provider duplicates are merged on Merges (IP, name, hostname) — one card, several sources. Changes have their own changelog. Details and screenshots: inventory docs.

Costs ​

Not the provider’s invoice — your money model:

  • Rate — what you will pay (monthly / yearly); the next due date is computed.
  • Charge — what you already paid; the next rate invoice can be marked paid.
  • Calendar — one-off events (domain, contract) plus a feed of upcoming payments.
  • By project — run-rate from rates; the project comes from the resource card.

Currencies are not summed into one number. The resource card shows the active rate. Details: costs.

Alerts ​

Rules over inventory and costs (agent offline, resource gone, amount threshold). Evaluation is about once a minute. A match opens an incident in the inbox; members acknowledge and resolve. Email, Slack, and anomaly detection are not in this version. Rule CRUD is owner-only. Details: alerts.

Agent ​

A process on your network: kadajgen/agent image, YAML config, local UI on port 8765. Plugins come from the catalog (complete today: Proxmox, Prometheus, SpaceWeb; Hetzner, Cloudflare, and AWS also install from Marketplace). Sync interval is per source; a failed plugin does not take the others down. Start: install → connect → plugins.

Team and access ​

Workspace, invites, workspace switch, profile, and console language (ru / en). The header has workspace search (Ctrl+K). API tokens are for the agent. Tenant data is isolated: each workspace has its own database. Details: team.

Getting started ​

  1. Sign in and accept a workspace invite (or create a workspace if the pilot allows it).
  2. Issue a token and install the agent.
  3. Wait for online under Agents.
  4. Sort the catalog, projects, and merges.
  5. Attach rates to resources and, if needed, alert rules.

Without an agent you can still fill the catalog by hand.