habakkuk serve is the long-running process. It runs the channels, the job workers (resuming interrupted turns), schedules, the board, background memory review, retention, and the HTTP server when there is an http section.

Run one serve per $HABAKKUK_HOME.
It stops cleanly on SIGINT or SIGTERM, and anything in flight resumes on the next start.
A small machine is enough: it is one static binary with SQLite storage, and the models run at your providers (or on a separate model server).

Pick a setup#

You have	Do this
A Linux or macOS machine that stays on	`habakkuk setup service` writes a systemd or launchd file; see Run it on a server
Docker	Run the image with a volume for `/data`; see In a container
A laptop you close at night	Run `serve` there anyway: work missed while it slept resumes, and late schedule slots are skipped
Webhook channels or a remote dashboard	Put it behind HTTPS; see Remote access

In this section#

Run it on a server systemd, launchd and Docker, and what serve needs.Remote access reverse proxies, tunnels and habakkuk expose tailscale.Security what an agent may do by default, the audit, and secret managers.Backups and updates habakkuk backup, restore, updates and retention.Observability status, health checks, logs, metrics and traces.