Lentago Labs is a shared learning lab, run by IT-operations people.
The estate is real — a Proxmox homelab cluster and a production-grade AWS platform — but the stakes are deliberately non-critical. Build it, break it, operate it, entirely through modern operations patterns, out in the open.
It's a place for the crew — IT-ops colleagues — to explore the estate, exercise the patterns, and carry the automation and agentic ideas back to the day job. Nothing here is claimed to be correct or unique — the patterns are compared against today's industry standards, on adoption and feature usage, to spot the ones that might be useful.
The lab · Real systems, survivable stakes, receipts in git.
The practice · Own your systems. Exit-ready by construction.
Modern operations, sized for mission-driven organizations.
The lab tests patterns in the open on real systems with survivable stakes; the practice puts the ones that earn their keep to work for organizations that run on volunteers and donations. The emphatic free-tier discipline is deliberate operating practice for exactly those constraints — not thrift.
The pledge — We will never host your systems for you. You'll own every piece, we'll teach your people, and firing us is a runbook.
DeepWiki maintains an AI-generated wiki over every public Lentago Labs repo — architecture pages, diagrams, and a Q&A box grounded in the actual code. It's the fastest way to orient before reading source. It is AI-generated: trust it to orient you, verify against the code before you act on it.
New to how any of this maps to enterprise practice? The asclepias glossary translates enterprise practice — CAB, CMDB, PIR and the rest — into what the lab actually does. It's the fastest orientation for either audience.
Everything is code. Every change is a pull request. Merges apply automatically. A self-hosted Claude agent fleet does directed work. Humans own every merge. That is the whole operating model — nothing here changes except through a reviewed PR, and the merged PR is the change record.
This is not a sandbox of toy YAML. Merge a PR in one of these repos and a live surface moves:
| Merge here… | …and it moves |
|---|---|
| drosera | your Grafana Cloud dashboards and alerts |
| kalmia | every Proxmox VM and LXC in the homelab |
| claytonia | the agent runner pool itself |
| solidago | the AWS platform |
| site repos | the live sites |
| .github | repo settings and rulesets, via fleet-ops |
Emphatically free-tier, wherever possible. When a service offers a free tier, that's the one the lab runs — caps and retention windows are treated as real operating constraints to be managed, not something to buy past.
Cloud & containers
Bare metal & virtualization
Infrastructure as code
CI/CD & supply chain
Observability & on-call
Each system splits a platform-agnostic core from per-source clients — the current build is always the first client, a working reference rather than a finished product.
 solidagoDeepWiki ↗ |
Reference three-tier AWS platform — 100% Terraform: VPC, ECS Fargate, RDS, WAF. |
 droseraDeepWiki ↗ |
Git-driven observability into Grafana Cloud — one Alloy container, Terraform-provisioned dashboards. |
 kalmiaDeepWiki ↗ |
Idempotent provisioning for workstations, VMs, and containers. |
 claytoniaDeepWiki ↗ |
Self-hosted agent fleet — drop a job, get a reviewed PR back. |
 betulaDeepWiki ↗ |
Full-volume log capture & archive → Axiom, at zero query cost. |
 asclepiasDeepWiki ↗ |
The field guide — operations manual, day-one path, hands-on labs. |
📖 Every public repo is indexed on DeepWiki — browse the wikis or ask the codebases anything.
Three ways in, depending on why you're here.
Taking part (the crew — this is what the lab is for)
- Pick a product repo above and read its 🛠️ Make a change yourself section.
- Ask that repo's DeepWiki a question about how it works.
- Run a lab — they ladder from ask the fleet up to owning a pattern.
- Open your first PR — small is fine; the required checks will guide you.
- Mention
@claudeon any issue or PR and watch the agent fleet respond.
Evaluating the operator (technical)
- solidago — the reference AWS platform, 100% Terraform.
- claytonia — the self-hosted agent fleet that does the directed work.
- homeassistant-config — physical-world monitoring, version-controlled.
Then the incident register for how things break, and the field-guide glossary to map it all onto enterprise practice.
Evaluating the practice (mission-driven organizations)
- The pledge (above) — you own every piece, we teach your people, firing us is a runbook.
- site-pondviewlane-com — a community organization's public-record fact base with a grounded Ask box.
- homeassistant-config — physical-world monitoring on a donated-scale budget.
- Incident register — post-mortems published verbatim.
Regenerated weekly from the repos themselves — we operate in the open.
- Fleet report — open issues by repo, a 7-day activity snapshot, and a code census that counts the
CLAUDE.md-family instruction files as natural-language code. - Language census — the canonical all-languages breakdown.
- Incident register — post-mortems from lab operations, with what broke, what did not, and the governance lessons.