Notebook
These short entries focus on operational clarity rather than novelty: clear restart paths, simple ingress, and documentation that survives personnel changes.
Single-node systems fail most often at the boundaries: ingress ownership, backup locations, and restart procedures that only exist in terminal history. The cheapest hardening step is to document where the service listens, which process owns each port, and how to restore the last known-good configuration from a local backup.
Configuration that can be explained in one screen usually survives handoffs better than layered automation. Fewer public listeners, explicit certificate paths, and plain-text runbooks are often more valuable than clever deployment wrappers.
A lightweight static site is enough for project notes, policy pages, and service verification. It avoids application drift, reduces update overhead, and still presents a normal HTTPS surface for browsers and search crawlers.