Open source from W&S Technik

Tools our in-house dev team built for itself — and now shares.

2 public projects

W&S Technik GmbH

Since 2007

W&S Technik is a service provider for railway infrastructure: we maintain signalling and safety systems, train specialist staff through the W&S Academy and build custom training installations — Technik, Consulting and Services from one source.

Founded2007
Employees100+
HeadquartersCastrop-Rauxel
BranchFriedrichsdorf
FieldRailway infrastructure
Technik Consulting Services wus-technik.com (opens in new tab)

Dev team

In-house software

A small in-house team that builds the software W&S colleagues use every day — internal platforms, automation and infrastructure. When a tool solves a problem others have too, we publish it under an open licence.

Public projects2 (MIT)
FocusInternal tools, DevOps, infrastructure

Claude Usage Tray

Windows tray app
Claude Usage Tray popup showing the 5-hour and 7-day usage bars with reset countdowns, a per-model weekly cap and monthly credit spend

Your Claude limits, always visible — right next to the clock.

MIT Windows 10/11 x64 .NET 10 No admin Auto-update

Claude Code tells you that you hit a limit after you hit it. This sits in your tray and tells you it’s coming — the rolling windows, the per-model weekly cap that throttles first, and how much paid credit you’ve burned this month.

  • Two badge iconsFill shows usage, colour shows severity, the digit names the window — 5 hours or 7 days.
  • Per-model capsWeekly limits scoped to a model or surface — the cap that usually bites first.
  • Credit spendPaid extra usage against your monthly limit, in your account’s currency.
  • Reset countdowns“Resets in 3h 58m” on hover and in the popup.
  • Pace marker & coloursA line shows where the clock is in the period; colour follows usage against the clock, not raw percent.
  • Platform statusA warning badge appears while status.claude.com reports anything but “All Systems Operational”.

Per-user install, nothing written outside your profile. Polls Anthropic’s usage API every 5 minutes and falls back to Claude Code’s local cache.

MeshCentral Agent for Docker

Container image

Put a container on your MeshCentral map — two variables and a volume.

MIT Docker Debian 13 Probe + slim GHCR

MeshCentral hands you an install script meant for a host, not a container. This image wraps it: point it at your server and a device group, mount one volume, and the agent installs itself once and comes straight back up on every restart. The probe variant brings a network toolkit, so a remote terminal on site is somewhere you can actually diagnose from.

  • Nothing baked inThe agent comes from your own server’s installer, so it always matches the server it reports to.
  • One volume/meshagent holds the installation — later starts find it and skip the download.
  • Stable identityPin the MAC address and the device keeps its NodeID across restarts; the container warns if it moves.
  • Probe or slimThe probe adds ~25 network tools — dig, mtr, tcpdump, nmap — for MeshCentral’s remote terminal; slim is the agent alone.
  • Fails loudlyA bad URL, group ID or installer stops the container with a message naming the cause, instead of half-installing.
  • Two release tracks:stable moves only on a tagged release, :latest follows every build; both leave an immutable tag to pin.
services:
  meshcentral-agent:
    image: ghcr.io/wus-technik/docker-meshcentral-agent:stable
    restart: unless-stopped
    hostname: meshagent-site01         # the device's name — set it once
    mac_address: "02:57:53:03:01:01"   # pins the NodeID
    volumes:
      - ./meshagent-data:/meshagent
    environment:
      MESH_SERVER_URL: "https://your-meshcentral-server.com"
      MESH_GROUP_ID: "your-group-id"   # write each $ as $$

Not affiliated with MeshCentral. The agent itself is downloaded from your own server and carries its own licence.