Built from operations · For bare metal operators

Bare metal monitoring built by people who run bare metal

We spent a decade operating servers across 60+ global locations. Now we build the monitoring tools we always needed. Lightweight, opinionated, open source.

Glassmkr takes its name from the glassmaking tradition. Bohemian crystal is impossibly complex to craft yet produces objects for everyday life. A drinking glass, a vase, a chandelier. Some pieces are ordinary with millions of copies. Some are singular masterpieces. All require deep craft.

We apply the same thinking to server monitoring. Beautiful products, complicated to engineer, needed every day.

01

Built from operations, not theory

Glassmkr comes from a decade of running bare metal infrastructure. Every alert, every threshold, every diagnostic comes from real operational experience, not market research.

02

Encounter first, then build

We only build features from real operational pain. If we have not hit the problem ourselves while running servers, we do not build for it.

03

Transparency by default

Open-source collector and MCP tools. Public documentation. Honest about what works and what does not. No black boxes in your monitoring stack.

04

Craft over speed

If anyone can build it in a day, it is not a product. We invest in understanding failure modes deeply enough to build monitoring that actually helps, not just dashboards that look pretty.

05

For operators, by operators

Our customers are people who manage servers. Hosting companies, MSPs, sysadmins, homelabs. We speak your language because we have been running the same infrastructure.

06

Free tools, paid reliability

Core monitoring is free and open-source. Run it yourself, alert yourself, no limits. Forge adds history, fleet views, and managed alerting for teams that want a dashboard without running one.

Live

Forge

Receives health data from your servers, stores history, displays fleet health, sends alerts when hardware or OS issues need attention.

Hardware health via IPMI/iDRAC/iLO
Storage monitoring with SMART and RAID awareness
Network interface diagnostics
OS alerts: RAM, swap, disk, CPU iowait, OOM kills
Telegram, email, and Slack alerting
AI-powered health analysis (Pro)
Free for up to 3 servers
Learn More →
Live
Open Source

Bench

MCP servers for infrastructure. Give AI agents structured access to Netdata, IPMI, Proxmox, and more. Each tool does one thing well.

38+ tools across 3 MCP servers
Read tools execute freely, write tools require confirmation
Works with Claude Code, Cursor, any MCP client
TypeScript, npm install, zero config
MIT licensed, fully open source
Browse Tools →