Operate in Practice
Small systems, visible operations, and proof from the site itself.
Proof from the Workbench
HappiHacking is small enough that its own infrastructure matters. The site, API, course material, hidden tools, deployment scripts, and agent-assisted workflows are part of the same operating practice described in the writing.
This page keeps the public version intentionally anonymized. It names the shape of the system and the operating discipline without publishing host details, secrets, network paths, or private runbooks.
Current Operating Shape
Small K3s Environment
A private K3s node, known internally as gds08, is used for staging and practice around deployments, service boundaries, health checks, and operational inspection.
Bahnhof Production Host
The public site runs on a Bahnhof VPS with Traefik, Docker, nginx for static content, and a small API container. Static deploys and API deploys are handled separately.
Agent-Assisted Delivery
Agents help with code reading, content cleanup, issue tracking, security checks, deployment preparation, and verification. Changes still need human review, build output, and operational checks.
Operating Habits
Clean Deploy Inputs
Static deploys are built from the intended site tree. Drafts, generated experiments, and unrelated workspace changes are kept out of production builds.
Separated Static and API Paths
The static site can be synced without touching the API. API deploys use a smoke-tested path with blue-green replacement on the production host.
Health Checks Before Claims
Deploys are followed by direct checks against representative pages, API health endpoints, containers, and canonical URLs.
Security Findings Become Work Items
CVE and package checks are turned into tracked tasks when the fix needs scheduling, verification, or a server change outside ordinary site edits.
Runtime Curiosity Stays Visible
The shell, CRT mode, N.I.S.S.E., and hidden utilities stay in the site because they make the operating culture inspectable.
Agents Leave Evidence
Agent work is useful when it leaves diffs, command output, issue references, and a path for a human to verify the result.
How This Connects to Advisory Work
The same habits show up in client work: understand the actual system, keep deployment and rollback paths boring, connect decisions to evidence, and make the next operator's job easier.
Small systems are useful practice grounds. They expose the same questions as larger systems: who owns state, what changes safely, where the logs point, how agents are constrained, and what happens when the day is already busy.
