CLI flags

Every flag per binary.

Every binary uses Go’s standard flag package. --help on any of them prints the full list.

Manager (cmd/manager)

The controller + scaler.

FlagDefaultNotes
--metrics-bind-address:8080Prometheus scrape target
--health-probe-bind-address:8081/healthz, /readyz
--leader-electfalseRequired when running multiple replicas
--enable-webhookstrueSet false to skip webhook server (Helm chart does)
--proxy-imageghcr.io/kryptonhq/krypton-proxy:latestSidecar image injected into agent pods
--enable-scalertrueDisable to run controller only
--scaler-interval-ms1000Tick interval for the scaling decider
--scaler-stable-window-ms60000Hysteresis: refuse scale-down within this window
--sidecar-port8888Where the scaler probes /_krypton/inflight

Control plane (cmd/control-plane)

REST API + UI.

FlagDefaultNotes
--api-bind-address:8090Public REST API + UI
--metrics-bind-address:8091Prometheus
--health-probe-bind-address:8092controller-runtime probes
--database-url$DATABASE_URLPostgres DSN; empty = in-memory store

Gateway (cmd/gateway)

Public ingress + activator.

FlagDefaultNotes
--listen-address:8080Public HTTP
--metrics-bind-address:8081Prometheus
--health-probe-bind-address:8082controller-runtime probes
--max-buffer-per-agent100Cold-start waiters per agent before 503
--poll-interval-ms50Endpoint-readiness poll interval
--default-startup-timeout-ms30000Used when spec.startupTimeout is unset

Sidecar (cmd/krypton-proxy)

Config via environment variables (the controller injects these at pod template render time):

Env varDefaultNotes
KRYPTON_AGENT_NAMEunknownMetric label
KRYPTON_AGENT_NAMESPACEdefaultMetric label
KRYPTON_LISTEN_ADDR:8888Sidecar port (Service TargetPort)
KRYPTON_UPSTREAM_URLhttp://127.0.0.1:8080User container address
KRYPTON_CONCURRENCY8In-flight cap; over → 503 + Retry-After
KRYPTON_MODEserverlessserverless or always-on
KRYPTON_IDLE_TIMEOUT300sIdle threshold (informational; scaler handles scale-to-zero)
KRYPTON_SHUTDOWN_TIMEOUT25sGraceful drain bound on SIGTERM