Merge pull request 'dev-deploy: production mirror + full observability behind the /_gm gate' (#88) from feature/dev-prod-mirror into development

- ARCHITECTURE §17: the dev (production-mirror) collection stack
  (Prometheus / Loki / Tempo / promtail / node-exporter / cAdvisor) and
  the single /_gm Basic Auth gate fronting Grafana and the Mailpit UI.
- tools/dev-deploy/monitoring/README.md (new): services, what is
  collected, Grafana-behind-the-gate access, config delivery, tuning.
- tools/dev-deploy/README.md: an Observability section; the Mailpit UI
  under /_gm/mailpit/; Networking diagram and Files list updated.
- FUNCTIONAL §10.2.1 (+ ru mirror): the operator console nav links to
  Grafana and Mailpit under the same /_gm gate, one sign-in for all.

.gitea/workflows/dev-deploy.yaml

@@ -148,12 +148,37 @@ jobs:
             -v "${{ gitea.workspace }}/pkg/geoip/test-data/test-data:/src:ro" \
             alpine sh -c 'cp /src/GeoIP2-Country-Test.mmdb /dst/geoip.mmdb'
 
+      - name: Seed mailpit relay config
+        env:
+          GALAXY_DEV_MAIL_RELAY_USERNAME: ${{ secrets.GALAXY_DEV_MAIL_RELAY_USERNAME }}
+          GALAXY_DEV_MAIL_RELAY_PASSWORD: ${{ secrets.GALAXY_DEV_MAIL_RELAY_PASSWORD }}
+        run: |
+          # Render the Mailpit relay upstream config from the template,
+          # substituting the Gmail App Password from a Gitea secret, then
+          # seed it into a named volume (same rationale as the geoip seed:
+          # a workspace bind-mount would vanish with the runner workspace).
+          # The secret never lands in git or a committed file; it is
+          # rendered to a tmpfile outside the repo and removed after. Gmail
+          # App Passwords are [a-z]{16}, so the `|` sed delimiter is safe.
+          # When the secret is unset the creds render empty and the compose
+          # default relay-match is non-routable, so the stack only captures.
+          rendered="$(mktemp)"
+          sed -e "s|\${GALAXY_DEV_MAIL_RELAY_USERNAME}|${GALAXY_DEV_MAIL_RELAY_USERNAME}|g" \
+              -e "s|\${GALAXY_DEV_MAIL_RELAY_PASSWORD}|${GALAXY_DEV_MAIL_RELAY_PASSWORD}|g" \
+              "${{ gitea.workspace }}/tools/dev-deploy/mailpit/relay.conf.tmpl" > "$rendered"
+          docker volume create galaxy-dev-mailpit-config >/dev/null
+          docker run --rm \
+            -v galaxy-dev-mailpit-config:/dst \
+            -v "$rendered:/src/relay.conf:ro" \
+            alpine sh -c 'cp /src/relay.conf /dst/relay.conf && chmod 600 /dst/relay.conf'
+          rm -f "$rendered"
+
       - name: Recycle engine containers on image drift
         run: |
           # Compare the freshly-built `galaxy-engine:dev` SHA against
           # every running `galaxy-game-*` container. The backend
           # reconciler adopts pre-existing labelled engine containers
-          # without checking image drift, so a running sandbox would
+          # without checking image drift, so a running game would
           # otherwise keep serving the previous engine code until the
           # container is recycled by hand. This step makes the recycle
           # automatic but only when it is actually needed:
@@ -168,10 +193,7 @@ jobs:
           #     silent state corruption otherwise), and cascade-delete
           #     the lobby `games` row (the FKs in `00001_init.sql`
           #     drop the matching `runtime_records`, `memberships`,
-          #     `player_mappings`, etc. in the same write). The
+          #     `player_mappings`, etc. in the same write).
-          #     `dev-sandbox` bootstrap on the next backend boot finds
-          #     no live sandbox and provisions a fresh one on the new
-          #     engine image.
           #
           # Backend is stopped first to keep the reconciler from
           # racing the recycle (mid-stream adoption / restart). The
@@ -234,11 +256,24 @@ jobs:
 
       - name: Bring up the stack
         working-directory: tools/dev-deploy
+        env:
+          # Recipient regex Mailpit auto-relays to the owner's Gmail.
+          # Unset/empty → the compose default (non-routable) keeps the
+          # stack capture-only.
+          GALAXY_DEV_MAIL_RELAY_MATCH: ${{ vars.GALAXY_DEV_MAIL_RELAY_MATCH }}
+          # Grafana admin password; unset/empty -> compose default 'admin'.
+          GALAXY_DEV_GRAFANA_ADMIN_PASSWORD: ${{ secrets.GALAXY_DEV_GRAFANA_ADMIN_PASSWORD }}
         run: |
           # Resolve in the shell, not in YAML expressions — `env.HOME`
           # is empty at the workflow-evaluation stage.
           export GALAXY_DEV_GAME_STATE_DIR="$HOME/.galaxy-dev/game-state"
           mkdir -p "$GALAXY_DEV_GAME_STATE_DIR"
+          # Seed the monitoring config to a stable, reboot-surviving host
+          # path (compose binds \${GALAXY_DEV_MONITORING_DIR} read-only).
+          export GALAXY_DEV_MONITORING_DIR="$HOME/.galaxy-dev/monitoring"
+          rm -rf "$GALAXY_DEV_MONITORING_DIR"
+          mkdir -p "$GALAXY_DEV_MONITORING_DIR"
+          cp -r monitoring/. "$GALAXY_DEV_MONITORING_DIR/"
           docker compose up -d --wait --remove-orphans
 
       - name: Probe the stack

backend/README.md

@@ -27,10 +27,16 @@ The implementation specification lives in `PLAN.md`.
 | ------------------ | ----------------------------------------------- | ------------------------------------- |
 | `/api/v1/public/*` | none                                            | Registration, code confirmation       |
 | `/api/v1/user/*`   | `X-User-ID` injected by gateway                 | Authenticated end users               |
-| `/api/v1/admin/*`  | HTTP Basic Auth against `admin_accounts`        | Platform administrators               |
+| `/api/v1/admin/*`  | HTTP Basic Auth against `admin_accounts`        | Platform administrators (JSON)        |
+| `/_gm`, `/_gm/*`   | HTTP Basic Auth against `admin_accounts`        | Operator console (server-rendered HTML)|
 | `/healthz`         | none                                            | Liveness probe                        |
 | `/readyz`          | none                                            | Readiness probe                       |
 
+The `/_gm` operator console is the human-facing surface for the admin
+operations; it reuses the admin Basic Auth verifier, renders with
+`html/template`, and is the only admin surface exposed publicly (through
+the gateway). See `docs/admin-console.md`.
+
 The full contract is documented in `openapi.yaml` and validated at
 runtime by the contract tests under `internal/server/`.
 
@@ -100,6 +106,7 @@ fast.
 | `BACKEND_GAME_STATE_ROOT`               | yes      | —                             | Host directory bind-mounted into engine containers.   |
 | `BACKEND_ADMIN_BOOTSTRAP_USER`          | no       | —                             | Initial admin username; idempotent insert.            |
 | `BACKEND_ADMIN_BOOTSTRAP_PASSWORD`      | no       | —                             | Initial admin password; required if user is set.      |
+| `BACKEND_ADMIN_CONSOLE_CSRF_KEY`        | no       | random per-process            | Secret keying the `/_gm` console CSRF token. Set a shared value across replicas; unset uses a per-process random key (forms reset on restart). |
 | `BACKEND_GEOIP_DB_PATH`                 | yes      | —                             | Filesystem path to GeoLite2 Country `.mmdb`.          |
 | `BACKEND_OTEL_TRACES_EXPORTER`          | no       | `otlp`                        | `none`, `otlp`, `stdout`.                             |
 | `BACKEND_OTEL_METRICS_EXPORTER`         | no       | `otlp`                        | `none`, `otlp`, `stdout`, `prometheus`.               |

backend/cmd/backend/main.go

@@ -22,10 +22,10 @@ import (
 	_ "time/tzdata"
 
 	"galaxy/backend/internal/admin"
+	"galaxy/backend/internal/adminconsole"
 	"galaxy/backend/internal/app"
 	"galaxy/backend/internal/auth"
 	"galaxy/backend/internal/config"
-	"galaxy/backend/internal/devsandbox"
 	"galaxy/backend/internal/diplomail"
 	"galaxy/backend/internal/diplomail/detector"
 	"galaxy/backend/internal/diplomail/translator"
@@ -37,6 +37,7 @@ import (
 	"galaxy/backend/internal/mail"
 	"galaxy/backend/internal/metricsapi"
 	"galaxy/backend/internal/notification"
+	"galaxy/backend/internal/opsstatus"
 	backendpostgres "galaxy/backend/internal/postgres"
 	"galaxy/backend/push"
 	"galaxy/backend/internal/runtime"
@@ -272,29 +273,18 @@ func run(ctx context.Context) (err error) {
 	)
 	runtimeGateway.svc = runtimeSvc
 
-	// Run a single reconciliation pass before the dev-sandbox
+	// Run a single reconciliation pass at startup so any runtime row
-	// bootstrap so any runtime row pointing at a vanished engine
+	// pointing at a vanished engine container (a host reboot wiped
-	// container (host reboot wiped /tmp/galaxy-game-state/<uuid>;
+	// /tmp/galaxy-game-state/<uuid>; `tools/local-dev`'s
-	// `tools/local-dev`'s `prune-broken-engines` target reaped the
+	// `prune-broken-engines` target reaped the husk) is cascaded
-	// husk) is already cascaded through `markRemoved` → lobby
+	// through `markRemoved` → lobby `cancelled` before the server
-	// `cancelled` by the time the bootstrap walks the sandbox list.
+	// starts serving requests. Failures are
-	// Without this pre-tick the bootstrap would reuse the
-	// soon-to-be-cancelled game and force the developer into a
-	// second `make up` cycle to land a healthy sandbox. Failures are
 	// non-fatal: the periodic ticker started later catches up, and
 	// the worst case degrades to the legacy two-cycle recovery.
 	if err := runtimeSvc.Reconciler().Tick(ctx); err != nil {
 		logger.Warn("pre-bootstrap reconciler tick failed", zap.Error(err))
 	}
 
-	if err := devsandbox.Bootstrap(ctx, devsandbox.Deps{
-		Users:          userSvc,
-		Lobby:          lobbySvc,
-		EngineVersions: engineVersionSvc,
-	}, cfg.DevSandbox, logger); err != nil {
-		return fmt.Errorf("dev sandbox bootstrap: %w", err)
-	}
-
 	notifStore := notification.NewStore(db)
 	notifSvc := notification.NewService(notification.Deps{
 		Store:    notifStore,
@@ -360,6 +350,32 @@ func run(ctx context.Context) (err error) {
 		return authCache.Ready() && userCache.Ready() && adminCache.Ready() && lobbyCache.Ready() && runtimeCache.Ready()
 	}
 
+	var consoleCSRF *adminconsole.CSRF
+	if cfg.AdminConsole.CSRFKey != "" {
+		consoleCSRF = adminconsole.NewCSRF([]byte(cfg.AdminConsole.CSRFKey))
+	} else {
+		consoleCSRF, err = adminconsole.NewRandomCSRF()
+		if err != nil {
+			return fmt.Errorf("init admin console CSRF: %w", err)
+		}
+		logger.Warn("admin console CSRF key not set; using a per-process random key (forms reset on restart, not valid across replicas)",
+			zap.String("env", "BACKEND_ADMIN_CONSOLE_CSRF_KEY"))
+	}
+	adminConsoleHandlers := backendserver.NewAdminConsoleHandlers(backendserver.AdminConsoleDeps{
+		CSRF:    consoleCSRF,
+		Monitor:        opsstatus.NewStore(db),
+		Ready:          ready,
+		Users:          userSvc,
+		Games:          lobbySvc,
+		Runtime:        runtimeSvc,
+		EngineVersions: engineVersionSvc,
+		Operators:      adminSvc,
+		Mail:           mailSvc,
+		Notifications:  notifSvc,
+		Diplomail:      diplomailSvc,
+		Logger:         logger,
+	})
+
 	handler, err := backendserver.NewRouter(backendserver.RouterDependencies{
 		Logger:                logger,
 		Telemetry:             telemetryRT,
@@ -388,6 +404,7 @@ func run(ctx context.Context) (err error) {
 		AdminGeo:              adminGeoHandlers,
 		UserGames:             userGamesHandlers,
 		UserMail:              userMailHandlers,
+		AdminConsole:          adminConsoleHandlers,
 	})
 	if err != nil {
 		return fmt.Errorf("build backend router: %w", err)

backend/docs/admin-console.md

@@ -0,0 +1,128 @@
+# Operator console (`/_gm`)
+
+The operator console is a server-rendered web UI for the platform's admin
+operations. It is the human-facing counterpart to the JSON admin API under
+`/api/v1/admin/*`: both call the same service layer, but the console renders
+HTML pages an operator drives in a browser, while the JSON API stays internal
+to the deployment for programmatic and test use.
+
+## Design choices
+
+- **Server-rendered, no client framework.** Pages are rendered with the
+  standard library's `html/template`. Navigation is by ordinary links and
+  query parameters; every state change is an HTML form `POST` answered with a
+  Post/Redirect/Get redirect. There is no build step, no JavaScript framework,
+  and no separate asset pipeline — a single embedded stylesheet under
+  `/_gm/assets/`.
+- **Reuses the existing admin auth.** The console mounts behind the same
+  `basicauth.Middleware(admin.Service)` verifier that gates `/api/v1/admin/*`,
+  so there is one credential store (`admin_accounts`, bcrypt-12) and no second
+  secret to manage.
+- **Lives in the backend.** The backend owns the admin domain and the data, so
+  rendering there lets the console call the service layer directly. The gateway
+  stays a thin proxy.
+
+## Request path
+
+```
+Browser ── /_gm/* ──► edge Caddy ──► gateway (public listener)
+   gateway: anti-abuse `admin` class (per-IP rate limit, body + method limits)
+            └─► reverse proxy ──► backend  /_gm/*
+   backend: basicauth.Middleware(admin.Service)
+            └─► CSRF guard (state-changing methods)
+                └─► console handler ──► admin service layer ──► html/template
+```
+
+The gateway preserves the inbound `Host` and relays the backend's `401` Basic
+Auth challenge unchanged, so the browser shows its native credential dialog.
+The gateway adds only the edge anti-abuse layer; authentication and every state
+change are enforced by the backend. The gateway answers `502` when the backend
+is unreachable. See the gateway README "Operator Console Proxy" section for the
+`admin` route-class env vars.
+
+## Components (package `internal/adminconsole`)
+
+The package is framework-agnostic (no gin) so it unit-tests in isolation:
+
+- `Renderer` — parses the embedded layout plus one content page per route and
+  renders a named page wrapped in the shared layout. Rendering goes through an
+  intermediate buffer, so a template failure never emits a partial document.
+- `CSRF` — issues and verifies the stateless anti-CSRF token: HMAC-SHA256 over
+  the authenticated username, keyed by `BACKEND_ADMIN_CONSOLE_CSRF_KEY`. When
+  the key is unset a per-process random key is used (secure, but forms reset on
+  restart and do not validate across replicas — set a shared key for
+  multi-replica deployments).
+- `Assets` — the embedded stylesheet filesystem served under `/_gm/assets/`.
+
+The gin glue (route group, Basic Auth, the CSRF guard middleware, the per-page
+handlers) lives in `internal/server/handlers_admin_console.go` and
+`internal/server/router.go` (`registerAdminConsoleRoutes`).
+
+## CSRF protection
+
+Because the console is sessionless (HTTP Basic Auth, whose credentials the
+browser replays automatically), state-changing requests are double-guarded:
+
+1. A stateless per-operator token (`_csrf` form field) that a cross-site page
+   cannot read or forge.
+2. A same-origin `Origin`/`Referer` check (when the browser sends one), which
+   relies on the gateway preserving the inbound `Host`.
+
+Safe methods (`GET`/`HEAD`/`OPTIONS`) pass without a token.
+
+## Monitoring
+
+The dashboard is the console landing page. It surfaces backend-visible
+operational state — service health, game-runtime status, and queue depths —
+read through the existing service and persistence layers. Richer cross-service
+metrics are out of scope for the console itself: the `/metrics` Prometheus
+exporters on `backend` and `gateway` are wired and enabled in the dev
+deployment so a future Prometheus + Grafana stack can scrape them without code
+changes.
+
+## Pages
+
+| Path                              | Method   | Purpose                                                        |
+| --------------------------------- | -------- | -------------------------------------------------------------- |
+| `/_gm`, `/_gm/`                   | GET      | Dashboard: health, runtime/mail/notification status, queues.   |
+| `/_gm/assets/*`                   | GET      | Embedded stylesheet.                                           |
+| `/_gm/users`                      | GET      | Paginated account list.                                        |
+| `/_gm/users/{id}`                 | GET      | Account detail: profile, entitlement, active sanctions.        |
+| `/_gm/users/{id}/block`           | POST     | Apply a permanent block (reason required).                     |
+| `/_gm/users/{id}/entitlement`     | POST     | Set the entitlement tier.                                      |
+| `/_gm/users/{id}/soft-delete`     | POST     | Soft-delete the account (cascades).                            |
+| `/_gm/games`                      | GET/POST | Paginated game list; POST creates a public game.               |
+| `/_gm/games/{id}`                 | GET      | Game detail with the runtime snapshot.                         |
+| `/_gm/games/{id}/force-start`     | POST     | Force-start the game.                                          |
+| `/_gm/games/{id}/force-stop`      | POST     | Force-stop the game.                                          |
+| `/_gm/games/{id}/ban-member`      | POST     | Ban a member (user id + reason).                               |
+| `/_gm/games/{id}/runtime/restart` | POST     | Restart the engine container.                                  |
+| `/_gm/games/{id}/runtime/patch`   | POST     | Patch the runtime to a target version.                         |
+| `/_gm/games/{id}/runtime/force-next-turn` | POST | Force the next turn now.                                  |
+| `/_gm/engine-versions`            | GET/POST | Version registry; POST registers a version.                    |
+| `/_gm/engine-versions/{ver}/disable` | POST  | Disable a registered version.                                  |
+| `/_gm/operators`                  | GET/POST | Admin-account list; POST creates an operator.                  |
+| `/_gm/operators/{user}/disable`   | POST     | Disable an operator.                                          |
+| `/_gm/operators/{user}/enable`    | POST     | Re-enable an operator.                                        |
+| `/_gm/operators/{user}/reset-password` | POST | Reset an operator's password.                                  |
+| `/_gm/mail`                       | GET      | Mail deliveries (paginated) + a dead-letter snapshot.          |
+| `/_gm/mail/deliveries/{id}`       | GET      | Delivery detail with its attempts.                             |
+| `/_gm/mail/deliveries/{id}/resend`| POST     | Re-enqueue a non-sent delivery.                                |
+| `/_gm/notifications`              | GET      | Notifications, dead-letters, and malformed intents overview.   |
+| `/_gm/broadcast`                  | GET/POST | Admin multi-game diplomatic broadcast.                         |
+
+Each page reuses the same service layer as the corresponding `/api/v1/admin/*`
+JSON endpoint; the console adds no business logic. Collection-mutating POSTs are
+mounted on the collection path (`POST /_gm/games`, `POST /_gm/engine-versions`)
+so a static action segment never collides with a path parameter in the gin
+router. Unblocking a user is not yet available because the JSON admin API
+exposes no remove-sanction endpoint.
+
+## Configuration
+
+| Variable                          | Where   | Notes                                                        |
+| --------------------------------- | ------- | ------------------------------------------------------------ |
+| `BACKEND_ADMIN_CONSOLE_CSRF_KEY`  | backend | CSRF token key; unset → per-process random key.              |
+| `BACKEND_ADMIN_BOOTSTRAP_USER`    | backend | Bootstrap operator account (shared with the JSON admin API). |
+| `BACKEND_ADMIN_BOOTSTRAP_PASSWORD`| backend | Bootstrap operator password.                                 |
+| `GATEWAY_PUBLIC_HTTP_ANTI_ABUSE_ADMIN_*` | gateway | `admin` route-class rate-limit and body budgets.      |

backend/internal/adminconsole/assets/console.css

@@ -0,0 +1,103 @@
+/* Admin console stylesheet. Deliberately small and dependency-free: the
+   console is an internal operator tool, not a public surface. */
+:root {
+  --bg: #11151c;
+  --panel: #1b2230;
+  --panel-hi: #232c3d;
+  --ink: #e6ebf2;
+  --ink-dim: #9aa7ba;
+  --line: #2c3850;
+  --accent: #5aa9ff;
+  --danger: #ff6b6b;
+  --ok: #4ecb8d;
+}
+* { box-sizing: border-box; }
+body {
+  margin: 0;
+  background: var(--bg);
+  color: var(--ink);
+  font: 15px/1.5 ui-monospace, SFMono-Regular, Menlo, Consolas, monospace;
+}
+a { color: var(--accent); text-decoration: none; }
+a:hover { text-decoration: underline; }
+.topbar {
+  display: flex;
+  align-items: center;
+  gap: 1.5rem;
+  padding: 0.6rem 1.2rem;
+  background: var(--panel);
+  border-bottom: 1px solid var(--line);
+}
+.topbar .brand { font-weight: 700; letter-spacing: 0.04em; }
+.topbar .mainnav { display: flex; gap: 1rem; flex: 1; }
+.topbar .mainnav a.active { color: var(--ink); border-bottom: 2px solid var(--accent); }
+.topbar .who { color: var(--ink-dim); }
+.content { padding: 1.5rem; max-width: 1100px; margin: 0 auto; }
+h1 { font-size: 1.4rem; margin: 0 0 0.4rem; }
+.lede { color: var(--ink-dim); margin-top: 0; }
+.cards { display: grid; grid-template-columns: repeat(auto-fit, minmax(220px, 1fr)); gap: 1rem; margin-top: 1.5rem; }
+.card {
+  display: block;
+  padding: 1rem 1.2rem;
+  background: var(--panel);
+  border: 1px solid var(--line);
+  border-radius: 8px;
+  color: var(--ink);
+}
+.card:hover { background: var(--panel-hi); text-decoration: none; }
+.card h2 { font-size: 1.05rem; margin: 0 0 0.3rem; color: var(--accent); }
+.card p { margin: 0; color: var(--ink-dim); font-size: 0.9rem; }
+
+.panel {
+  padding: 0.9rem 1.1rem;
+  background: var(--panel);
+  border: 1px solid var(--line);
+  border-radius: 8px;
+  margin-bottom: 1rem;
+}
+.panel h2 { font-size: 1rem; margin: 0 0 0.6rem; color: var(--ink); }
+.grid { display: grid; grid-template-columns: repeat(auto-fit, minmax(220px, 1fr)); gap: 1rem; margin-bottom: 1rem; }
+.grid .panel { margin-bottom: 0; }
+.kv { list-style: none; margin: 0; padding: 0; }
+.kv li { padding: 0.15rem 0; color: var(--ink-dim); }
+.counts { width: 100%; border-collapse: collapse; font-size: 0.9rem; }
+.counts td { padding: 0.2rem 0; border-bottom: 1px solid var(--line); color: var(--ink-dim); }
+.counts td.num { text-align: right; color: var(--ink); font-variant-numeric: tabular-nums; }
+.bignum { font-size: 1.6rem; margin: 0; color: var(--ink); }
+.note { color: var(--ink-dim); font-style: italic; margin: 0.2rem 0; }
+.errors { border-color: var(--danger); }
+.errors ul { margin: 0; padding-left: 1.1rem; color: var(--danger); }
+.ok { color: var(--ok); }
+.bad { color: var(--danger); }
+
+.list { width: 100%; border-collapse: collapse; font-size: 0.9rem; margin-bottom: 1rem; }
+.list th, .list td { text-align: left; padding: 0.35rem 0.6rem; border-bottom: 1px solid var(--line); }
+.list th { color: var(--ink-dim); font-weight: 600; }
+.list tr:hover td { background: var(--panel-hi); }
+.pager { display: flex; gap: 1rem; align-items: center; color: var(--ink-dim); }
+.form { display: flex; flex-wrap: wrap; gap: 0.6rem; align-items: end; margin-top: 0.8rem; }
+.form label { display: flex; flex-direction: column; gap: 0.2rem; font-size: 0.85rem; color: var(--ink-dim); }
+.form input, .form select {
+  background: var(--bg);
+  color: var(--ink);
+  border: 1px solid var(--line);
+  border-radius: 6px;
+  padding: 0.35rem 0.5rem;
+  font: inherit;
+}
+button {
+  background: var(--accent);
+  color: #06121f;
+  border: 0;
+  border-radius: 6px;
+  padding: 0.4rem 0.9rem;
+  font: inherit;
+  font-weight: 600;
+  cursor: pointer;
+}
+button:hover { filter: brightness(1.1); }
+button.danger { background: var(--danger); color: #1a0606; }
+code { background: var(--bg); padding: 0.05rem 0.3rem; border-radius: 4px; }
+.actions { display: flex; flex-wrap: wrap; gap: 0.6rem; margin: 0.8rem 0; }
+.actions form { margin: 0; }
+.subnav { color: var(--ink-dim); margin: -0.3rem 0 1rem; font-size: 0.9rem; }

backend/internal/adminconsole/csrf.go

@@ -0,0 +1,54 @@
+package adminconsole
+
+import (
+	"crypto/hmac"
+	"crypto/rand"
+	"crypto/sha256"
+	"encoding/base64"
+	"fmt"
+)
+
+// CSRF issues and verifies the stateless anti-CSRF token used by the admin
+// console. The token is an HMAC-SHA256 over the authenticated operator's
+// username keyed by a process secret, so a cross-site request cannot forge it
+// without already being able to read an authenticated page. The console is
+// sessionless (HTTP Basic Auth), which makes a stateless, per-operator token
+// the natural fit.
+type CSRF struct {
+	key []byte
+}
+
+// NewCSRF returns a CSRF signer keyed by key. A shared key across backend
+// replicas lets a form rendered by one replica validate on another; callers
+// that pass a per-process random key (see NewRandomCSRF) accept that forms do
+// not survive a restart or span replicas.
+func NewCSRF(key []byte) *CSRF {
+	return &CSRF{key: key}
+}
+
+// NewRandomCSRF returns a CSRF signer keyed by a fresh 32-byte random secret.
+// It is the secure default when no shared key is configured.
+func NewRandomCSRF() (*CSRF, error) {
+	key := make([]byte, 32)
+	if _, err := rand.Read(key); err != nil {
+		return nil, fmt.Errorf("generate admin console CSRF key: %w", err)
+	}
+	return &CSRF{key: key}, nil
+}
+
+// Token returns the anti-CSRF token bound to username.
+func (c *CSRF) Token(username string) string {
+	mac := hmac.New(sha256.New, c.key)
+	mac.Write([]byte(username))
+	return base64.RawURLEncoding.EncodeToString(mac.Sum(nil))
+}
+
+// Verify reports whether token is the valid anti-CSRF token for username. The
+// comparison runs in constant time relative to the token bytes.
+func (c *CSRF) Verify(username, token string) bool {
+	if token == "" {
+		return false
+	}
+	expected := c.Token(username)
+	return hmac.Equal([]byte(token), []byte(expected))
+}

backend/internal/adminconsole/csrf_test.go

@@ -0,0 +1,42 @@
+package adminconsole
+
+import "testing"
+
+func TestCSRFTokenRoundTrip(t *testing.T) {
+	signer := NewCSRF([]byte("shared-secret"))
+	token := signer.Token("alice")
+
+	if !signer.Verify("alice", token) {
+		t.Fatal("valid token rejected")
+	}
+	if signer.Verify("bob", token) {
+		t.Fatal("token accepted for a different operator")
+	}
+	if signer.Verify("alice", "") {
+		t.Fatal("empty token accepted")
+	}
+	if signer.Verify("alice", token+"x") {
+		t.Fatal("tampered token accepted")
+	}
+}
+
+func TestCSRFKeySeparation(t *testing.T) {
+	a := NewCSRF([]byte("key-a"))
+	b := NewCSRF([]byte("key-b"))
+	if a.Token("operator") == b.Token("operator") {
+		t.Fatal("tokens collide across distinct keys")
+	}
+	if b.Verify("operator", a.Token("operator")) {
+		t.Fatal("token minted under one key verified under another")
+	}
+}
+
+func TestRandomCSRFRoundTrip(t *testing.T) {
+	signer, err := NewRandomCSRF()
+	if err != nil {
+		t.Fatalf("NewRandomCSRF: %v", err)
+	}
+	if !signer.Verify("operator", signer.Token("operator")) {
+		t.Fatal("random-key token failed to round-trip")
+	}
+}

backend/internal/adminconsole/dashboard.go

@@ -0,0 +1,24 @@
+package adminconsole
+
+// StatusCount pairs a status label with its current row count for the
+// dashboard's per-status tables. It is the view-layer counterpart of the
+// data gathered by the ops-status reader; the server handler maps between
+// them so this package stays free of database concerns.
+type StatusCount struct {
+	Status string
+	Count  int64
+}
+
+// DashboardData is the view model for the console landing page. MonitorAvailable
+// is false when no ops-status reader is wired, in which case the monitoring
+// panels are omitted. Errors carries non-fatal probe failures for display.
+type DashboardData struct {
+	MonitorAvailable      bool
+	BackendReady          bool
+	PostgresHealthy       bool
+	Runtimes              []StatusCount
+	MailDeliveries        []StatusCount
+	NotificationRoutes    []StatusCount
+	NotificationMalformed int64
+	Errors                []string
+}

backend/internal/adminconsole/doc.go

@@ -0,0 +1,18 @@
+// Package adminconsole renders the server-side operator console mounted by the
+// backend under the `/_gm` route group.
+//
+// The console is a multi-page, server-rendered surface built on the standard
+// library's html/template package: navigation is driven by request path and
+// query, state changes are submitted with HTML forms and answered with a
+// Post/Redirect/Get redirect. The package owns three concerns and nothing
+// transport-specific:
+//
+//   - Renderer composes the shared layout with one content page per route.
+//   - CSRF issues and verifies the stateless anti-CSRF token embedded in every
+//     state-changing form.
+//   - Assets exposes the embedded stylesheet served under `/_gm/assets/`.
+//
+// The gin glue (route registration, Basic Auth, the CSRF guard middleware, and
+// the per-page handlers) lives in package server; this package stays free of
+// the web framework so it can be unit-tested in isolation.
+package adminconsole

backend/internal/adminconsole/games.go

@@ -0,0 +1,67 @@
+package adminconsole
+
+// GameRow is one line in the games list table.
+type GameRow struct {
+	GameID       string
+	GameName     string
+	Visibility   string
+	Status       string
+	Owner        string
+	Players      string
+	TurnSchedule string
+	CreatedAt    string
+}
+
+// GamesListData is the view model for the paginated games list.
+type GamesListData struct {
+	Items    []GameRow
+	Page     int
+	PageSize int
+	Total    int
+	HasPrev  bool
+	HasNext  bool
+	PrevPage int
+	NextPage int
+}
+
+// GameDetailData is the view model for a single game, combining the lobby
+// record with the runtime snapshot and the available actions.
+type GameDetailData struct {
+	GameID              string
+	GameName            string
+	Description         string
+	Visibility          string
+	Status              string
+	Owner               string
+	MinPlayers          int32
+	MaxPlayers          int32
+	StartGapHours       int32
+	StartGapPlayers     int32
+	TurnSchedule        string
+	TargetEngineVersion string
+	EnrollmentEndsAt    string
+	CreatedAt           string
+	StartedAt           string
+	FinishedAt          string
+
+	HasRuntime           bool
+	RuntimeStatus        string
+	CurrentEngineVersion string
+	EngineHealth         string
+	CurrentTurn          int32
+	NextGenerationAt     string
+	Paused               bool
+}
+
+// EngineVersionRow is one line in the engine-version registry table.
+type EngineVersionRow struct {
+	Version   string
+	ImageRef  string
+	Enabled   bool
+	CreatedAt string
+}
+
+// EngineVersionsData is the view model for the engine-version registry page.
+type EngineVersionsData struct {
+	Items []EngineVersionRow
+}

backend/internal/adminconsole/mail.go

@@ -0,0 +1,86 @@
+package adminconsole
+
+// MailDeliveryRow is one line in the mail deliveries table.
+type MailDeliveryRow struct {
+	DeliveryID  string
+	Template    string
+	Status      string
+	Attempts    int32
+	NextAttempt string
+	Created     string
+}
+
+// MailDeadLetterRow is one line in the mail dead-letters table.
+type MailDeadLetterRow struct {
+	DeliveryID string
+	Reason     string
+	Archived   string
+}
+
+// MailData is the view model for the mail page: a paginated deliveries list
+// plus a snapshot of dead-letters.
+type MailData struct {
+	Deliveries  []MailDeliveryRow
+	DeadLetters []MailDeadLetterRow
+	Page        int
+	PageSize    int
+	Total       int64
+	HasPrev     bool
+	HasNext     bool
+	PrevPage    int
+	NextPage    int
+}
+
+// MailAttemptRow is one delivery attempt on the mail detail page.
+type MailAttemptRow struct {
+	AttemptNo int32
+	Outcome   string
+	Started   string
+	Finished  string
+	Error     string
+}
+
+// MailDeliveryDetail is the view model for a single delivery.
+type MailDeliveryDetail struct {
+	DeliveryID   string
+	Template     string
+	Status       string
+	Attempts     int32
+	NextAttempt  string
+	LastError    string
+	Created      string
+	Sent         string
+	DeadLettered string
+	CanResend    bool
+	AttemptRows  []MailAttemptRow
+}
+
+// NotificationRow is one line in the notifications table.
+type NotificationRow struct {
+	NotificationID string
+	Kind           string
+	UserID         string
+	Created        string
+}
+
+// NotificationDeadLetterRow is one line in the notification dead-letters table.
+type NotificationDeadLetterRow struct {
+	NotificationID string
+	RouteID        string
+	Reason         string
+	Archived       string
+}
+
+// MalformedRow is one line in the malformed-intents table.
+type MalformedRow struct {
+	ID       string
+	Reason   string
+	Received string
+}
+
+// NotificationsData is the view model for the notifications overview page.
+type NotificationsData struct {
+	Notifications []NotificationRow
+	DeadLetters   []NotificationDeadLetterRow
+	Malformed     []MalformedRow
+}

backend/internal/adminconsole/message.go

@@ -0,0 +1,11 @@
+package adminconsole
+
+// MessageData is the view model for the generic message page used to render
+// not-found, validation, and operation-failure notices. Class selects the CSS
+// styling (for example "bad" for errors); BackHref, when set, renders a link
+// back to a relevant page.
+type MessageData struct {
+	Message  string
+	Class    string
+	BackHref string
+}

backend/internal/adminconsole/operators.go

@@ -0,0 +1,14 @@
+package adminconsole
+
+// OperatorRow is one line in the operators (admin accounts) table.
+type OperatorRow struct {
+	Username   string
+	CreatedAt  string
+	LastUsedAt string
+	Disabled   bool
+}
+
+// OperatorsData is the view model for the operators page.
+type OperatorsData struct {
+	Items []OperatorRow
+}

backend/internal/adminconsole/render.go

@@ -0,0 +1,107 @@
+package adminconsole
+
+import (
+	"bytes"
+	"embed"
+	"fmt"
+	"html/template"
+	"io"
+	"io/fs"
+	"path"
+	"strings"
+)
+
+//go:embed templates
+var templatesFS embed.FS
+
+//go:embed assets
+var assetsFS embed.FS
+
+// Renderer holds the parsed admin console templates. It composes one template
+// set per content page, each combining the shared layout (defining the page
+// chrome and the "layout" entry template) with that page's "content" block, so
+// rendering a page is a single ExecuteTemplate call against the "layout" name.
+type Renderer struct {
+	pages map[string]*template.Template
+}
+
+// PageData is the view model passed to every admin console page. Title is the
+// document title; Username is the authenticated operator; CSRFToken is the
+// per-operator token embedded into state-changing forms; ActiveNav marks the
+// highlighted navigation entry; Data carries the page-specific payload.
+type PageData struct {
+	Title     string
+	Username  string
+	CSRFToken string
+	ActiveNav string
+	Data      any
+}
+
+// NewRenderer parses the embedded layout and every content page under
+// templates/pages, returning a Renderer ready to serve them. It fails when a
+// template cannot be parsed.
+func NewRenderer() (*Renderer, error) {
+	base, err := template.New("layout").ParseFS(templatesFS, "templates/layout.gohtml")
+	if err != nil {
+		return nil, fmt.Errorf("parse admin console layout: %w", err)
+	}
+
+	pageFiles, err := fs.Glob(templatesFS, "templates/pages/*.gohtml")
+	if err != nil {
+		return nil, fmt.Errorf("enumerate admin console pages: %w", err)
+	}
+	if len(pageFiles) == 0 {
+		return nil, fmt.Errorf("admin console: no page templates found under templates/pages")
+	}
+
+	pages := make(map[string]*template.Template, len(pageFiles))
+	for _, file := range pageFiles {
+		name := strings.TrimSuffix(path.Base(file), ".gohtml")
+		clone, err := base.Clone()
+		if err != nil {
+			return nil, fmt.Errorf("clone admin console layout for %q: %w", name, err)
+		}
+		if _, err := clone.ParseFS(templatesFS, file); err != nil {
+			return nil, fmt.Errorf("parse admin console page %q: %w", name, err)
+		}
+		pages[name] = clone
+	}
+
+	return &Renderer{pages: pages}, nil
+}
+
+// MustNewRenderer is like NewRenderer but panics on error. The templates are
+// embedded at build time, so a parse failure is a programmer error rather than
+// a runtime condition.
+func MustNewRenderer() *Renderer {
+	renderer, err := NewRenderer()
+	if err != nil {
+		panic(err)
+	}
+	return renderer
+}
+
+// Render writes the named page, wrapped in the shared layout, to w using data.
+// It returns an error when page is unknown or template execution fails; the
+// page is rendered into an intermediate buffer first so a mid-render failure
+// never emits a partial document to w.
+func (r *Renderer) Render(w io.Writer, page string, data PageData) error {
+	tmpl, ok := r.pages[page]
+	if !ok {
+		return fmt.Errorf("admin console: unknown page %q", page)
+	}
+
+	var buf bytes.Buffer
+	if err := tmpl.ExecuteTemplate(&buf, "layout", data); err != nil {
+		return fmt.Errorf("render admin console page %q: %w", page, err)
+	}
+
+	_, err := buf.WriteTo(w)
+	return err
+}
+
+// Assets returns the embedded static asset tree rooted at the assets directory,
+// suitable for serving under `/_gm/assets/`.
+func Assets() (fs.FS, error) {
+	return fs.Sub(assetsFS, "assets")
+}

backend/internal/adminconsole/render_test.go

@@ -0,0 +1,67 @@
+package adminconsole
+
+import (
+	"bytes"
+	"io/fs"
+	"strings"
+	"testing"
+)
+
+func TestRendererRendersDashboard(t *testing.T) {
+	renderer, err := NewRenderer()
+	if err != nil {
+		t.Fatalf("NewRenderer: %v", err)
+	}
+
+	var buf bytes.Buffer
+	err = renderer.Render(&buf, "dashboard", PageData{
+		Title:     "Dashboard",
+		Username:  "ops-bob",
+		ActiveNav: "dashboard",
+	})
+	if err != nil {
+		t.Fatalf("Render: %v", err)
+	}
+
+	out := buf.String()
+	for _, want := range []string{
+		"<!DOCTYPE html>",
+		"Dashboard",
+		"ops-bob",
+		`href="/_gm/users"`,
+		"/_gm/assets/console.css",
+	} {
+		if !strings.Contains(out, want) {
+			t.Errorf("rendered page missing %q\n--- page ---\n%s", want, out)
+		}
+	}
+}
+
+func TestRendererUnknownPage(t *testing.T) {
+	renderer := MustNewRenderer()
+	if err := renderer.Render(&bytes.Buffer{}, "does-not-exist", PageData{}); err == nil {
+		t.Fatal("expected an error rendering an unknown page")
+	}
+}
+
+func TestRendererEscapesUsername(t *testing.T) {
+	renderer := MustNewRenderer()
+
+	var buf bytes.Buffer
+	if err := renderer.Render(&buf, "dashboard", PageData{Username: "<script>evil</script>"}); err != nil {
+		t.Fatalf("Render: %v", err)
+	}
+	if strings.Contains(buf.String(), "<script>evil</script>") {
+		t.Error("username was not HTML-escaped in the rendered page")
+	}
+}
+
+func TestAssetsContainsStylesheet(t *testing.T) {
+	fsys, err := Assets()
+	if err != nil {
+		t.Fatalf("Assets: %v", err)
+	}
+	if _, err := fs.Stat(fsys, "console.css"); err != nil {
+		t.Fatalf("console.css missing from embedded assets: %v", err)
+	}
+}

backend/internal/adminconsole/templates/layout.gohtml

@@ -0,0 +1,30 @@
+{{define "layout" -}}
+<!DOCTYPE html>
+<html lang="en">
+<head>
+<meta charset="utf-8">
+<meta name="viewport" content="width=device-width, initial-scale=1">
+<meta name="robots" content="noindex, nofollow">
+<title>{{.Title}} · Galaxy GM</title>
+<link rel="stylesheet" href="/_gm/assets/console.css">
+</head>
+<body>
+<header class="topbar">
+  <span class="brand">Galaxy · GM</span>
+  <nav class="mainnav">
+    <a href="/_gm/"{{if eq .ActiveNav "dashboard"}} class="active"{{end}}>Dashboard</a>
+    <a href="/_gm/users"{{if eq .ActiveNav "users"}} class="active"{{end}}>Users</a>
+    <a href="/_gm/games"{{if eq .ActiveNav "games"}} class="active"{{end}}>Games</a>
+    <a href="/_gm/operators"{{if eq .ActiveNav "operators"}} class="active"{{end}}>Operators</a>
+    <a href="/_gm/mail"{{if eq .ActiveNav "mail"}} class="active"{{end}}>Mail</a>
+    <a href="/_gm/grafana/" target="_blank" rel="noopener">Grafana</a>
+    <a href="/_gm/mailpit/" target="_blank" rel="noopener">Mailpit</a>
+  </nav>
+  <span class="who">{{.Username}}</span>
+</header>
+<main class="content">
+{{template "content" .}}
+</main>
+</body>
+</html>
+{{- end}}

backend/internal/adminconsole/templates/pages/broadcast.gohtml

@@ -0,0 +1,21 @@
+{{define "content" -}}
+{{$csrf := .CSRFToken}}
+<h1>Broadcast</h1>
+<nav class="subnav"><a href="/_gm/mail">Deliveries</a> · <a href="/_gm/notifications">Notifications</a> · <a href="/_gm/broadcast">Broadcast</a></nav>
+<section class="panel">
+<h2>Admin broadcast</h2>
+<form method="post" action="/_gm/broadcast" class="form">
+<input type="hidden" name="_csrf" value="{{$csrf}}">
+<label>Scope
+<select name="scope"><option value="all_running">all running games</option><option value="selected">selected games</option></select>
+</label>
+<label>Game IDs (comma-separated, for "selected") <input type="text" name="game_ids" placeholder="uuid,uuid"></label>
+<label>Recipients
+<select name="recipients"><option value="active">active members</option><option value="active_and_removed">active and removed</option><option value="all_members">all members</option></select>
+</label>
+<label>Subject <input type="text" name="subject"></label>
+<label>Body <input type="text" name="body" required></label>
+<button type="submit">Send broadcast</button>
+</form>
+</section>
+{{- end}}

backend/internal/adminconsole/templates/pages/dashboard.gohtml

@@ -0,0 +1,69 @@
+{{define "content" -}}
+<h1>Dashboard</h1>
+<p class="lede">Signed in as <strong>{{.Username}}</strong>.</p>
+{{with .Data}}
+<section class="panel">
+  <h2>Health</h2>
+  <ul class="kv">
+    <li>Backend ready: {{if .BackendReady}}<span class="ok">yes</span>{{else}}<span class="bad">no</span>{{end}}</li>
+    <li>Postgres: {{if .PostgresHealthy}}<span class="ok">healthy</span>{{else}}<span class="bad">unreachable</span>{{end}}</li>
+  </ul>
+</section>
+{{if .MonitorAvailable}}
+<div class="grid">
+  <section class="panel">
+    <h2>Game runtimes</h2>
+    {{template "statuscounts" .Runtimes}}
+  </section>
+  <section class="panel">
+    <h2>Mail deliveries</h2>
+    {{template "statuscounts" .MailDeliveries}}
+  </section>
+  <section class="panel">
+    <h2>Notification routes</h2>
+    {{template "statuscounts" .NotificationRoutes}}
+  </section>
+  <section class="panel">
+    <h2>Malformed notifications</h2>
+    <p class="bignum {{if gt .NotificationMalformed 0}}bad{{end}}">{{.NotificationMalformed}}</p>
+  </section>
+</div>
+{{if .Errors}}
+<section class="panel errors">
+  <h2>Collection errors</h2>
+  <ul>{{range .Errors}}<li>{{.}}</li>{{end}}</ul>
+</section>
+{{end}}
+{{else}}
+<p class="note">Monitoring is not wired in this deployment.</p>
+{{end}}
+{{end}}
+<section class="cards">
+  <a class="card" href="/_gm/users">
+    <h2>Users</h2>
+    <p>Accounts, sanctions, entitlements, soft-delete.</p>
+  </a>
+  <a class="card" href="/_gm/games">
+    <h2>Games &amp; runtimes</h2>
+    <p>Lobby state, engine versions, turn control.</p>
+  </a>
+  <a class="card" href="/_gm/operators">
+    <h2>Operators</h2>
+    <p>Admin accounts: create, disable, reset password.</p>
+  </a>
+  <a class="card" href="/_gm/mail">
+    <h2>Mail &amp; notifications</h2>
+    <p>Deliveries, dead-letters, broadcasts.</p>
+  </a>
+</section>
+{{- end}}
+
+{{define "statuscounts" -}}
+{{if .}}
+<table class="counts"><tbody>
+{{range .}}<tr><td>{{.Status}}</td><td class="num">{{.Count}}</td></tr>{{end}}
+</tbody></table>
+{{else}}
+<p class="note">none</p>
+{{end}}
+{{- end}}

backend/internal/adminconsole/templates/pages/engine_versions.gohtml

@@ -0,0 +1,30 @@
+{{define "content" -}}
+{{$csrf := .CSRFToken}}
+<h1>Engine versions</h1>
+{{with .Data}}
+<table class="list">
+<thead><tr><th>Version</th><th>Image</th><th>Enabled</th><th>Created</th><th></th></tr></thead>
+<tbody>
+{{range .Items}}
+<tr>
+<td>{{.Version}}</td>
+<td><code>{{.ImageRef}}</code></td>
+<td>{{if .Enabled}}<span class="ok">yes</span>{{else}}<span class="bad">no</span>{{end}}</td>
+<td>{{.CreatedAt}}</td>
+<td>{{if .Enabled}}<form method="post" action="/_gm/engine-versions/{{.Version}}/disable" onsubmit="return confirm('Disable {{.Version}}?');"><input type="hidden" name="_csrf" value="{{$csrf}}"><button type="submit" class="danger">Disable</button></form>{{end}}</td>
+</tr>
+{{else}}<tr><td colspan="5"><span class="note">no engine versions</span></td></tr>{{end}}
+</tbody>
+</table>
+{{end}}
+<section class="panel">
+<h2>Register version</h2>
+<form method="post" action="/_gm/engine-versions" class="form">
+<input type="hidden" name="_csrf" value="{{$csrf}}">
+<label>Version <input type="text" name="version" placeholder="semver e.g. 0.1.0" required></label>
+<label>Image ref <input type="text" name="image_ref" required></label>
+<label>Enabled <input type="checkbox" name="enabled" value="true" checked></label>
+<button type="submit">Register</button>
+</form>
+</section>
+{{- end}}

backend/internal/adminconsole/templates/pages/game_detail.gohtml

@@ -0,0 +1,65 @@
+{{define "content" -}}
+{{$csrf := .CSRFToken}}
+{{with .Data}}
+<p><a href="/_gm/games">&laquo; all games</a></p>
+<h1>{{if .GameName}}{{.GameName}}{{else}}(unnamed){{end}}</h1>
+
+<section class="panel">
+<h2>Game</h2>
+<ul class="kv">
+<li>Game ID: <code>{{.GameID}}</code></li>
+<li>Visibility: {{.Visibility}}</li>
+<li>Status: {{.Status}}</li>
+<li>Owner: {{.Owner}}</li>
+<li>Players: {{.MinPlayers}}–{{.MaxPlayers}}</li>
+<li>Start gap: {{.StartGapHours}}h / {{.StartGapPlayers}} players</li>
+<li>Turn schedule: {{.TurnSchedule}}</li>
+<li>Target engine: {{.TargetEngineVersion}}</li>
+<li>Enrollment ends: {{.EnrollmentEndsAt}}</li>
+<li>Created: {{.CreatedAt}}</li>
+<li>Started: {{if .StartedAt}}{{.StartedAt}}{{else}}—{{end}}</li>
+<li>Finished: {{if .FinishedAt}}{{.FinishedAt}}{{else}}—{{end}}</li>
+</ul>
+{{if .Description}}<p>{{.Description}}</p>{{end}}
+<div class="actions">
+<form method="post" action="/_gm/games/{{.GameID}}/force-start" onsubmit="return confirm('Force-start this game?');"><input type="hidden" name="_csrf" value="{{$csrf}}"><button type="submit">Force start</button></form>
+<form method="post" action="/_gm/games/{{.GameID}}/force-stop" onsubmit="return confirm('Force-stop this game?');"><input type="hidden" name="_csrf" value="{{$csrf}}"><button type="submit" class="danger">Force stop</button></form>
+</div>
+</section>
+
+<section class="panel">
+<h2>Runtime</h2>
+{{if .HasRuntime}}
+<ul class="kv">
+<li>Status: {{.RuntimeStatus}}</li>
+<li>Engine version: {{.CurrentEngineVersion}}</li>
+<li>Engine health: {{.EngineHealth}}</li>
+<li>Current turn: {{.CurrentTurn}}</li>
+<li>Next generation: {{if .NextGenerationAt}}{{.NextGenerationAt}}{{else}}—{{end}}</li>
+<li>Paused: {{if .Paused}}yes{{else}}no{{end}}</li>
+</ul>
+<div class="actions">
+<form method="post" action="/_gm/games/{{.GameID}}/runtime/restart" onsubmit="return confirm('Restart the engine container?');"><input type="hidden" name="_csrf" value="{{$csrf}}"><button type="submit">Restart</button></form>
+<form method="post" action="/_gm/games/{{.GameID}}/runtime/force-next-turn" onsubmit="return confirm('Force the next turn now?');"><input type="hidden" name="_csrf" value="{{$csrf}}"><button type="submit">Force next turn</button></form>
+</div>
+<form method="post" action="/_gm/games/{{.GameID}}/runtime/patch" class="form">
+<input type="hidden" name="_csrf" value="{{$csrf}}">
+<label>Patch to version <input type="text" name="target_version" placeholder="e.g. 0.1.1" required></label>
+<button type="submit">Patch</button>
+</form>
+{{else}}
+<p class="note">No runtime record for this game yet.</p>
+{{end}}
+</section>
+
+<section class="panel">
+<h2>Ban member</h2>
+<form method="post" action="/_gm/games/{{.GameID}}/ban-member" class="form" onsubmit="return confirm('Ban this member from the game?');">
+<input type="hidden" name="_csrf" value="{{$csrf}}">
+<label>User ID <input type="text" name="user_id" required></label>
+<label>Reason <input type="text" name="reason"></label>
+<button type="submit" class="danger">Ban member</button>
+</form>
+</section>
+{{end}}
+{{- end}}

backend/internal/adminconsole/templates/pages/games.gohtml

@@ -0,0 +1,43 @@
+{{define "content" -}}
+{{$csrf := .CSRFToken}}
+<h1>Games</h1>
+{{with .Data}}
+<table class="list">
+<thead><tr><th>Name</th><th>Visibility</th><th>Status</th><th>Owner</th><th>Players</th><th>Schedule</th><th>Created</th></tr></thead>
+<tbody>
+{{range .Items}}
+<tr>
+<td><a href="/_gm/games/{{.GameID}}">{{if .GameName}}{{.GameName}}{{else}}(unnamed){{end}}</a></td>
+<td>{{.Visibility}}</td>
+<td>{{.Status}}</td>
+<td>{{.Owner}}</td>
+<td>{{.Players}}</td>
+<td>{{.TurnSchedule}}</td>
+<td>{{.CreatedAt}}</td>
+</tr>
+{{else}}<tr><td colspan="7"><span class="note">no games</span></td></tr>{{end}}
+</tbody>
+</table>
+<nav class="pager">
+{{if .HasPrev}}<a href="/_gm/games?page={{.PrevPage}}&amp;page_size={{.PageSize}}">&laquo; prev</a>{{end}}
+<span>page {{.Page}} · {{.Total}} total</span>
+{{if .HasNext}}<a href="/_gm/games?page={{.NextPage}}&amp;page_size={{.PageSize}}">next &raquo;</a>{{end}}
+</nav>
+{{end}}
+<section class="panel">
+<h2>Create public game</h2>
+<form method="post" action="/_gm/games" class="form">
+<input type="hidden" name="_csrf" value="{{$csrf}}">
+<label>Name <input type="text" name="game_name" required></label>
+<label>Description <input type="text" name="description"></label>
+<label>Min players <input type="number" name="min_players" value="2" min="1"></label>
+<label>Max players <input type="number" name="max_players" value="8" min="1"></label>
+<label>Start gap hours <input type="number" name="start_gap_hours" value="0" min="0"></label>
+<label>Start gap players <input type="number" name="start_gap_players" value="0" min="0"></label>
+<label>Enrollment ends <input type="datetime-local" name="enrollment_ends_at" required></label>
+<label>Turn schedule <input type="text" name="turn_schedule" placeholder="e.g. @every 24h" required></label>
+<label>Engine version <input type="text" name="target_engine_version" placeholder="e.g. 0.1.0" required></label>
+<button type="submit">Create</button>
+</form>
+</section>
+{{- end}}

backend/internal/adminconsole/templates/pages/mail.gohtml

@@ -0,0 +1,32 @@
+{{define "content" -}}
+<h1>Mail</h1>
+<nav class="subnav"><a href="/_gm/mail">Deliveries</a> · <a href="/_gm/notifications">Notifications</a> · <a href="/_gm/broadcast">Broadcast</a></nav>
+{{with .Data}}
+<section class="panel">
+<h2>Deliveries</h2>
+<table class="list">
+<thead><tr><th>Delivery</th><th>Template</th><th>Status</th><th>Attempts</th><th>Next attempt</th><th>Created</th></tr></thead>
+<tbody>
+{{range .Deliveries}}
+<tr><td><a href="/_gm/mail/deliveries/{{.DeliveryID}}"><code>{{.DeliveryID}}</code></a></td><td>{{.Template}}</td><td>{{.Status}}</td><td>{{.Attempts}}</td><td>{{if .NextAttempt}}{{.NextAttempt}}{{else}}—{{end}}</td><td>{{.Created}}</td></tr>
+{{else}}<tr><td colspan="6"><span class="note">no deliveries</span></td></tr>{{end}}
+</tbody>
+</table>
+<nav class="pager">
+{{if .HasPrev}}<a href="/_gm/mail?page={{.PrevPage}}&amp;page_size={{.PageSize}}">&laquo; prev</a>{{end}}
+<span>page {{.Page}} · {{.Total}} total</span>
+{{if .HasNext}}<a href="/_gm/mail?page={{.NextPage}}&amp;page_size={{.PageSize}}">next &raquo;</a>{{end}}
+</nav>
+</section>
+<section class="panel">
+<h2>Dead-letters</h2>
+<table class="list">
+<thead><tr><th>Delivery</th><th>Reason</th><th>Archived</th></tr></thead>
+<tbody>
+{{range .DeadLetters}}<tr><td><a href="/_gm/mail/deliveries/{{.DeliveryID}}"><code>{{.DeliveryID}}</code></a></td><td>{{.Reason}}</td><td>{{.Archived}}</td></tr>
+{{else}}<tr><td colspan="3"><span class="note">no dead-letters</span></td></tr>{{end}}
+</tbody>
+</table>
+</section>
+{{end}}
+{{- end}}

backend/internal/adminconsole/templates/pages/mail_delivery.gohtml

@@ -0,0 +1,33 @@
+{{define "content" -}}
+{{$csrf := .CSRFToken}}
+{{with .Data}}
+<p><a href="/_gm/mail">&laquo; mail</a></p>
+<h1>Delivery</h1>
+<section class="panel">
+<ul class="kv">
+<li>Delivery ID: <code>{{.DeliveryID}}</code></li>
+<li>Template: {{.Template}}</li>
+<li>Status: {{.Status}}</li>
+<li>Attempts: {{.Attempts}}</li>
+<li>Next attempt: {{if .NextAttempt}}{{.NextAttempt}}{{else}}—{{end}}</li>
+<li>Created: {{.Created}}</li>
+<li>Sent: {{if .Sent}}{{.Sent}}{{else}}—{{end}}</li>
+<li>Dead-lettered: {{if .DeadLettered}}{{.DeadLettered}}{{else}}—{{end}}</li>
+<li>Last error: {{if .LastError}}{{.LastError}}{{else}}—{{end}}</li>
+</ul>
+{{if .CanResend}}
+<form method="post" action="/_gm/mail/deliveries/{{.DeliveryID}}/resend" class="form" onsubmit="return confirm('Resend this delivery?');"><input type="hidden" name="_csrf" value="{{$csrf}}"><button type="submit">Resend</button></form>
+{{else}}<p class="note">Already sent — resend is not available.</p>{{end}}
+</section>
+<section class="panel">
+<h2>Attempts</h2>
+<table class="list">
+<thead><tr><th>#</th><th>Outcome</th><th>Started</th><th>Finished</th><th>Error</th></tr></thead>
+<tbody>
+{{range .AttemptRows}}<tr><td>{{.AttemptNo}}</td><td>{{.Outcome}}</td><td>{{.Started}}</td><td>{{if .Finished}}{{.Finished}}{{else}}—{{end}}</td><td>{{.Error}}</td></tr>
+{{else}}<tr><td colspan="5"><span class="note">no attempts</span></td></tr>{{end}}
+</tbody>
+</table>
+</section>
+{{end}}
+{{- end}}

backend/internal/adminconsole/templates/pages/message.gohtml

@@ -0,0 +1,7 @@
+{{define "content" -}}
+<h1>{{.Title}}</h1>
+{{with .Data}}
+<p class="{{.Class}}">{{.Message}}</p>
+{{if .BackHref}}<p><a href="{{.BackHref}}">&laquo; back</a></p>{{end}}
+{{end}}
+{{- end}}

backend/internal/adminconsole/templates/pages/notifications.gohtml

@@ -0,0 +1,27 @@
+{{define "content" -}}
+<h1>Notifications</h1>
+<nav class="subnav"><a href="/_gm/mail">Deliveries</a> · <a href="/_gm/notifications">Notifications</a> · <a href="/_gm/broadcast">Broadcast</a></nav>
+{{with .Data}}
+<section class="panel">
+<h2>Recent notifications</h2>
+<table class="list"><thead><tr><th>ID</th><th>Kind</th><th>User</th><th>Created</th></tr></thead><tbody>
+{{range .Notifications}}<tr><td><code>{{.NotificationID}}</code></td><td>{{.Kind}}</td><td>{{.UserID}}</td><td>{{.Created}}</td></tr>
+{{else}}<tr><td colspan="4"><span class="note">none</span></td></tr>{{end}}
+</tbody></table>
+</section>
+<section class="panel">
+<h2>Dead-letters</h2>
+<table class="list"><thead><tr><th>Notification</th><th>Route</th><th>Reason</th><th>Archived</th></tr></thead><tbody>
+{{range .DeadLetters}}<tr><td><code>{{.NotificationID}}</code></td><td><code>{{.RouteID}}</code></td><td>{{.Reason}}</td><td>{{.Archived}}</td></tr>
+{{else}}<tr><td colspan="4"><span class="note">none</span></td></tr>{{end}}
+</tbody></table>
+</section>
+<section class="panel">
+<h2>Malformed intents</h2>
+<table class="list"><thead><tr><th>ID</th><th>Reason</th><th>Received</th></tr></thead><tbody>
+{{range .Malformed}}<tr><td><code>{{.ID}}</code></td><td>{{.Reason}}</td><td>{{.Received}}</td></tr>
+{{else}}<tr><td colspan="3"><span class="note">none</span></td></tr>{{end}}
+</tbody></table>
+</section>
+{{end}}
+{{- end}}

backend/internal/adminconsole/templates/pages/operators.gohtml

@@ -0,0 +1,38 @@
+{{define "content" -}}
+{{$csrf := .CSRFToken}}
+<h1>Operators</h1>
+{{with .Data}}
+<table class="list">
+<thead><tr><th>Username</th><th>Status</th><th>Created</th><th>Last used</th><th>Actions</th></tr></thead>
+<tbody>
+{{range .Items}}
+<tr>
+<td>{{.Username}}</td>
+<td>{{if .Disabled}}<span class="bad">disabled</span>{{else}}<span class="ok">active</span>{{end}}</td>
+<td>{{.CreatedAt}}</td>
+<td>{{if .LastUsedAt}}{{.LastUsedAt}}{{else}}—{{end}}</td>
+<td>
+<div class="actions">
+{{if .Disabled}}
+<form method="post" action="/_gm/operators/{{.Username}}/enable"><input type="hidden" name="_csrf" value="{{$csrf}}"><button type="submit">Enable</button></form>
+{{else}}
+<form method="post" action="/_gm/operators/{{.Username}}/disable" onsubmit="return confirm('Disable {{.Username}}?');"><input type="hidden" name="_csrf" value="{{$csrf}}"><button type="submit" class="danger">Disable</button></form>
+{{end}}
+<form method="post" action="/_gm/operators/{{.Username}}/reset-password" class="form"><input type="hidden" name="_csrf" value="{{$csrf}}"><input type="password" name="password" placeholder="new password" required><button type="submit">Reset</button></form>
+</div>
+</td>
+</tr>
+{{else}}<tr><td colspan="5"><span class="note">no operators</span></td></tr>{{end}}
+</tbody>
+</table>
+{{end}}
+<section class="panel">
+<h2>Create operator</h2>
+<form method="post" action="/_gm/operators" class="form">
+<input type="hidden" name="_csrf" value="{{$csrf}}">
+<label>Username <input type="text" name="username" required></label>
+<label>Password <input type="password" name="password" required></label>
+<button type="submit">Create</button>
+</form>
+</section>
+{{- end}}

backend/internal/adminconsole/templates/pages/user_detail.gohtml

@@ -0,0 +1,68 @@
+{{define "content" -}}
+{{$csrf := .CSRFToken}}
+{{with .Data}}
+<p><a href="/_gm/users">&laquo; all users</a></p>
+<h1>{{.Email}}</h1>
+{{if .Deleted}}<p class="bad">This account is soft-deleted.</p>{{end}}
+
+<section class="panel">
+<h2>Account</h2>
+<ul class="kv">
+<li>User ID: <code>{{.UserID}}</code></li>
+<li>User name: {{.UserName}}</li>
+<li>Display name: {{.DisplayName}}</li>
+<li>Preferred language: {{.PreferredLanguage}}</li>
+<li>Time zone: {{.TimeZone}}</li>
+<li>Declared country: {{.DeclaredCountry}}</li>
+<li>Status: {{if .Blocked}}<span class="bad">blocked</span>{{else}}<span class="ok">active</span>{{end}}</li>
+<li>Created: {{.CreatedAt}}</li>
+<li>Updated: {{.UpdatedAt}}</li>
+</ul>
+</section>
+
+<section class="panel">
+<h2>Entitlement</h2>
+<ul class="kv">
+<li>Tier: <strong>{{.Tier}}</strong> ({{if .IsPaid}}paid{{else}}free{{end}})</li>
+<li>Source: {{.EntitlementSource}}</li>
+<li>Reason: {{.EntitlementReason}}</li>
+<li>Ends: {{if .EntitlementEnds}}{{.EntitlementEnds}}{{else}}—{{end}}</li>
+</ul>
+<form method="post" action="/_gm/users/{{.UserID}}/entitlement" class="form">
+<input type="hidden" name="_csrf" value="{{$csrf}}">
+<label>Tier
+<select name="tier">{{range .Tiers}}<option value="{{.}}">{{.}}</option>{{end}}</select>
+</label>
+<label>Source <input type="text" name="source" value="admin"></label>
+<label>Reason <input type="text" name="reason_code" placeholder="optional"></label>
+<button type="submit">Update entitlement</button>
+</form>
+</section>
+
+<section class="panel">
+<h2>Active sanctions</h2>
+{{if .Sanctions}}
+<table class="counts"><tbody>
+{{range .Sanctions}}<tr><td>{{.SanctionCode}}</td><td>{{.Scope}}</td><td>{{.ReasonCode}}</td><td>{{.AppliedAt}}</td></tr>{{end}}
+</tbody></table>
+{{else}}<p class="note">none</p>{{end}}
+{{if .Blocked}}
+<p class="note">User is permanently blocked. Unblock is not available in the current admin API.</p>
+{{else}}
+<form method="post" action="/_gm/users/{{.UserID}}/block" class="form" onsubmit="return confirm('Permanently block this user?');">
+<input type="hidden" name="_csrf" value="{{$csrf}}">
+<label>Reason <input type="text" name="reason_code" required></label>
+<button type="submit" class="danger">Permanently block</button>
+</form>
+{{end}}
+</section>
+
+<section class="panel">
+<h2>Danger zone</h2>
+<form method="post" action="/_gm/users/{{.UserID}}/soft-delete" class="form" onsubmit="return confirm('Soft-delete this account? This cascades to sessions, memberships, and owned games.');">
+<input type="hidden" name="_csrf" value="{{$csrf}}">
+<button type="submit" class="danger">Soft-delete account</button>
+</form>
+</section>
+{{end}}
+{{- end}}

backend/internal/adminconsole/templates/pages/users.gohtml

@@ -0,0 +1,27 @@
+{{define "content" -}}
+<h1>Users</h1>
+{{with .Data}}
+<table class="list">
+<thead><tr><th>Email</th><th>User name</th><th>Display</th><th>Tier</th><th>Status</th><th>Created</th></tr></thead>
+<tbody>
+{{range .Items}}
+<tr>
+<td><a href="/_gm/users/{{.UserID}}">{{.Email}}</a></td>
+<td>{{.UserName}}</td>
+<td>{{.DisplayName}}</td>
+<td>{{.Tier}}</td>
+<td>{{if .Deleted}}<span class="bad">deleted</span>{{else if .Blocked}}<span class="bad">blocked</span>{{else}}<span class="ok">active</span>{{end}}</td>
+<td>{{.CreatedAt}}</td>
+</tr>
+{{else}}
+<tr><td colspan="6"><span class="note">no users</span></td></tr>
+{{end}}
+</tbody>
+</table>
+<nav class="pager">
+{{if .HasPrev}}<a href="/_gm/users?page={{.PrevPage}}&amp;page_size={{.PageSize}}">&laquo; prev</a>{{end}}
+<span>page {{.Page}} · {{.Total}} total</span>
+{{if .HasNext}}<a href="/_gm/users?page={{.NextPage}}&amp;page_size={{.PageSize}}">next &raquo;</a>{{end}}
+</nav>
+{{end}}
+{{- end}}

backend/internal/adminconsole/users.go

@@ -0,0 +1,61 @@
+package adminconsole
+
+// UserRow is one line in the users list table.
+type UserRow struct {
+	UserID      string
+	Email       string
+	UserName    string
+	DisplayName string
+	Tier        string
+	Blocked     bool
+	Deleted     bool
+	CreatedAt   string
+}
+
+// UsersListData is the view model for the paginated users list.
+type UsersListData struct {
+	Items    []UserRow
+	Page     int
+	PageSize int
+	Total    int
+	HasPrev  bool
+	HasNext  bool
+	PrevPage int
+	NextPage int
+}
+
+// SanctionView is one active sanction shown on the user detail page.
+type SanctionView struct {
+	SanctionCode string
+	Scope        string
+	ReasonCode   string
+	AppliedAt    string
+	ExpiresAt    string
+}
+
+// UserDetailData is the view model for a single user's detail page,
+// combining the account aggregate with the form option lists.
+type UserDetailData struct {
+	UserID            string
+	Email             string
+	UserName          string
+	DisplayName       string
+	PreferredLanguage string
+	TimeZone          string
+	DeclaredCountry   string
+	Blocked           bool
+	Deleted           bool
+	CreatedAt         string
+	UpdatedAt         string
+
+	Tier              string
+	IsPaid            bool
+	EntitlementSource string
+	EntitlementReason string
+	EntitlementEnds   string
+
+	Sanctions []SanctionView
+
+	// Tiers lists the selectable entitlement tiers for the form.
+	Tiers []string
+}

backend/internal/config/config.go

@@ -55,6 +55,8 @@ const (
 	envAdminBootstrapUser = "BACKEND_ADMIN_BOOTSTRAP_USER"
 	envAdminBootstrapPassword = "BACKEND_ADMIN_BOOTSTRAP_PASSWORD"
 
+	envAdminConsoleCSRFKey = "BACKEND_ADMIN_CONSOLE_CSRF_KEY"
+
 	envGeoIPDBPath = "BACKEND_GEOIP_DB_PATH"
 
 	envOTelTracesExporter = "BACKEND_OTEL_TRACES_EXPORTER"
@@ -103,11 +105,6 @@ const (
 	envDiplomailTranslatorTimeout = "BACKEND_DIPLOMAIL_TRANSLATOR_TIMEOUT"
 	envDiplomailTranslatorMaxAttempts = "BACKEND_DIPLOMAIL_TRANSLATOR_MAX_ATTEMPTS"
 	envDiplomailWorkerInterval = "BACKEND_DIPLOMAIL_WORKER_INTERVAL"
-
-	envDevSandboxEmail = "BACKEND_DEV_SANDBOX_EMAIL"
-	envDevSandboxEngineImage = "BACKEND_DEV_SANDBOX_ENGINE_IMAGE"
-	envDevSandboxEngineVersion = "BACKEND_DEV_SANDBOX_ENGINE_VERSION"
-	envDevSandboxPlayerCount = "BACKEND_DEV_SANDBOX_PLAYER_COUNT"
 )
 
 // Default values applied when an environment variable is absent.
@@ -176,9 +173,6 @@ const (
 	defaultDiplomailTranslatorTimeout = 10 * time.Second
 	defaultDiplomailTranslatorMaxAttempts = 5
 	defaultDiplomailWorkerInterval = 2 * time.Second
-
-	defaultDevSandboxEngineVersion = "0.1.0"
-	defaultDevSandboxPlayerCount = 20
 )
 
 // Allowed values for the closed-set string options.
@@ -208,6 +202,7 @@ type Config struct {
 	Docker DockerConfig
 	Game GameConfig
 	Admin AdminBootstrapConfig
+	AdminConsole AdminConsoleConfig
 	GeoIP GeoIPConfig
 	Telemetry TelemetryConfig
 	Auth AuthConfig
@@ -216,29 +211,12 @@ type Config struct {
 	Runtime RuntimeConfig
 	Notification NotificationConfig
 	Diplomail DiplomailConfig
-	DevSandbox DevSandboxConfig
 
 	// FreshnessWindow mirrors the gateway freshness window and is used by the
 	// push server to bound the cursor TTL.
 	FreshnessWindow time.Duration
 }
 
-// DevSandboxConfig configures the boot-time bootstrap implemented in
-// `backend/internal/devsandbox`. When Email is empty the bootstrap
-// is a no-op, which is the production posture. When Email is set —
-// from `BACKEND_DEV_SANDBOX_EMAIL` in the `tools/local-dev` stack —
-// the bootstrap idempotently provisions a real user, the configured
-// number of dummy participants, a private "Dev Sandbox" game, the
-// matching memberships, and drives the lifecycle to `running`. The
-// engine image and engine version refer to a row that the bootstrap
-// also seeds in `engine_versions`.
-type DevSandboxConfig struct {
-	Email string
-	EngineImage string
-	EngineVersion string
-	PlayerCount int
-}
-
 // LoggingConfig stores the parameters used by the structured logger.
 type LoggingConfig struct {
 	// Level is the zap level name (e.g. "debug", "info", "warn", "error").
@@ -308,6 +286,15 @@ type AdminBootstrapConfig struct {
 	Password string
 }
 
+// AdminConsoleConfig configures the server-rendered operator console.
+// CSRFKey is the secret keying the console's stateless anti-CSRF token.
+// When empty the console falls back to a per-process random key, which is
+// secure but means forms do not survive a restart and do not validate across
+// replicas; set a shared key when running more than one backend instance.
+type AdminConsoleConfig struct {
+	CSRFKey string
+}
+
 // GeoIPConfig configures the GeoLite2 country database used by geo lookups.
 type GeoIPConfig struct {
 	DBPath string
@@ -560,10 +547,6 @@ func DefaultConfig() Config {
 			TranslatorMaxAttempts: defaultDiplomailTranslatorMaxAttempts,
 			WorkerInterval: defaultDiplomailWorkerInterval,
 		},
-		DevSandbox: DevSandboxConfig{
-			EngineVersion: defaultDevSandboxEngineVersion,
-			PlayerCount: defaultDevSandboxPlayerCount,
-		},
 		Runtime: RuntimeConfig{
 			WorkerPoolSize: defaultRuntimeWorkerPoolSize,
 			JobQueueSize: defaultRuntimeJobQueueSize,
@@ -644,6 +627,8 @@ func LoadFromEnv() (Config, error) {
 	cfg.Admin.User = loadString(envAdminBootstrapUser, cfg.Admin.User)
 	cfg.Admin.Password = loadString(envAdminBootstrapPassword, cfg.Admin.Password)
 
+	cfg.AdminConsole.CSRFKey = loadString(envAdminConsoleCSRFKey, cfg.AdminConsole.CSRFKey)
+
 	cfg.GeoIP.DBPath = loadString(envGeoIPDBPath, cfg.GeoIP.DBPath)
 
 	cfg.Telemetry.TracesExporter = strings.ToLower(loadString(envOTelTracesExporter, cfg.Telemetry.TracesExporter))
@@ -741,13 +726,6 @@ func LoadFromEnv() (Config, error) {
 		return Config{}, err
 	}
 
-	cfg.DevSandbox.Email = strings.TrimSpace(loadString(envDevSandboxEmail, cfg.DevSandbox.Email))
-	cfg.DevSandbox.EngineImage = strings.TrimSpace(loadString(envDevSandboxEngineImage, cfg.DevSandbox.EngineImage))
-	cfg.DevSandbox.EngineVersion = strings.TrimSpace(loadString(envDevSandboxEngineVersion, cfg.DevSandbox.EngineVersion))
-	if cfg.DevSandbox.PlayerCount, err = loadInt(envDevSandboxPlayerCount, cfg.DevSandbox.PlayerCount); err != nil {
-		return Config{}, err
-	}
-
 	if err := cfg.Validate(); err != nil {
 		return Config{}, err
 	}
@@ -959,21 +937,6 @@ func (c Config) Validate() error {
 		}
 	}
 
-	if email := strings.TrimSpace(c.DevSandbox.Email); email != "" {
-		if _, err := netmail.ParseAddress(email); err != nil {
-			return fmt.Errorf("%s must be a valid RFC 5322 address: %w", envDevSandboxEmail, err)
-		}
-		if strings.TrimSpace(c.DevSandbox.EngineImage) == "" {
-			return fmt.Errorf("%s must not be empty when %s is set", envDevSandboxEngineImage, envDevSandboxEmail)
-		}
-		if strings.TrimSpace(c.DevSandbox.EngineVersion) == "" {
-			return fmt.Errorf("%s must not be empty when %s is set", envDevSandboxEngineVersion, envDevSandboxEmail)
-		}
-		if c.DevSandbox.PlayerCount <= 0 {
-			return fmt.Errorf("%s must be positive when %s is set", envDevSandboxPlayerCount, envDevSandboxEmail)
-		}
-	}
-
 	return nil
 }
 

backend/internal/devsandbox/bootstrap.go

@@ -1,287 +0,0 @@
-// Package devsandbox provisions a ready-to-play game on backend boot
-// for the `tools/local-dev` stack.
-//
-// Bootstrap is invoked from `backend/cmd/backend/main.go` after the
-// admin bootstrap and before the HTTP listener starts. It reads
-// `cfg.DevSandbox`; when `Email` is empty (the production posture)
-// the function logs "skipped" and returns nil. When set, it
-// idempotently:
-//
-//   1. registers the configured engine version and image;
-//   2. find-or-creates the real dev user with the configured email;
-//   3. find-or-creates `cfg.PlayerCount - 1` deterministic dummy
-//      users so the engine's minimum-players constraint is met;
-//   4. find-or-creates a private "Dev Sandbox" game owned by the
-//      real user with min/max_players = cfg.PlayerCount and a
-//      year-out turn schedule (effectively frozen at turn 1);
-//   5. inserts memberships for all participants bypassing the
-//      application/approval flow;
-//   6. drives the lifecycle to `running` (or as far as possible if
-//      the runtime is busy).
-//
-// The function is a no-op on subsequent boots once the game is
-// running; partial states from earlier crashes are recovered.
-package devsandbox
-
-import (
-	"context"
-	"errors"
-	"fmt"
-	"time"
-
-	"galaxy/backend/internal/config"
-	"galaxy/backend/internal/lobby"
-	"galaxy/backend/internal/runtime"
-
-	"github.com/google/uuid"
-	"go.uber.org/zap"
-)
-
-// SandboxGameName is the display name used to identify the
-// auto-provisioned game on subsequent reboots. The combination of
-// game_name and owner_user_id is unique enough in practice — only
-// the dev sandbox bootstrap creates a game owned by the configured
-// real user with this exact name.
-const SandboxGameName = "Dev Sandbox"
-
-// SandboxTurnSchedule keeps the game on turn 1 by scheduling the
-// next turn a year out. The runtime scheduler still parses this and
-// will tick once a year — long enough to never interfere with
-// solo UI development.
-const SandboxTurnSchedule = "0 0 1 1 *"
-
-// UserEnsurer matches `auth.UserEnsurer`. We define a local
-// interface to avoid importing the auth package and circular
-// dependencies — the production wiring passes the same `*user.Service`
-// instance used by auth.
-type UserEnsurer interface {
-	EnsureByEmail(ctx context.Context, email, preferredLanguage, timeZone, declaredCountry string) (uuid.UUID, error)
-}
-
-// Deps aggregates the collaborators Bootstrap needs.
-type Deps struct {
-	Users          UserEnsurer
-	Lobby          *lobby.Service
-	EngineVersions *runtime.EngineVersionService
-}
-
-// Bootstrap runs the seven-step provisioning flow described on the
-// package doc comment. Errors are returned to the caller; the boot
-// path in `cmd/backend/main.go` aborts startup if Bootstrap fails so
-// a misconfigured dev environment surfaces immediately rather than
-// silently leaving the lobby empty.
-func Bootstrap(ctx context.Context, deps Deps, cfg config.DevSandboxConfig, logger *zap.Logger) error {
-	if logger == nil {
-		logger = zap.NewNop()
-	}
-	logger = logger.Named("dev_sandbox")
-
-	if cfg.Email == "" {
-		logger.Info("skipped (no email)")
-		return nil
-	}
-	if deps.Users == nil || deps.Lobby == nil || deps.EngineVersions == nil {
-		return errors.New("dev_sandbox: deps.Users, deps.Lobby and deps.EngineVersions are required")
-	}
-	if cfg.PlayerCount <= 0 {
-		return fmt.Errorf("dev_sandbox: PlayerCount must be positive, got %d", cfg.PlayerCount)
-	}
-
-	if err := ensureEngineVersion(ctx, deps.EngineVersions, cfg, logger); err != nil {
-		return err
-	}
-
-	realID, err := deps.Users.EnsureByEmail(ctx, cfg.Email, "en", "UTC", "")
-	if err != nil {
-		return fmt.Errorf("dev_sandbox: ensure real user: %w", err)
-	}
-
-	dummyIDs := make([]uuid.UUID, 0, cfg.PlayerCount-1)
-	for i := 1; i < cfg.PlayerCount; i++ {
-		email := fmt.Sprintf("dev-dummy-%02d@local.test", i)
-		id, err := deps.Users.EnsureByEmail(ctx, email, "en", "UTC", "")
-		if err != nil {
-			return fmt.Errorf("dev_sandbox: ensure dummy %d: %w", i, err)
-		}
-		dummyIDs = append(dummyIDs, id)
-	}
-
-	if err := purgeTerminalSandboxGames(ctx, deps.Lobby, realID, logger); err != nil {
-		return err
-	}
-
-	game, err := findOrCreateSandboxGame(ctx, deps.Lobby, realID, cfg)
-	if err != nil {
-		return err
-	}
-
-	game, err = ensureMembershipsAndDrive(ctx, deps.Lobby, game, realID, dummyIDs, logger)
-	if err != nil {
-		return err
-	}
-
-	logger.Info("bootstrap complete",
-		zap.String("user_id", realID.String()),
-		zap.String("game_id", game.GameID.String()),
-		zap.String("status", game.Status),
-	)
-	return nil
-}
-
-func ensureEngineVersion(ctx context.Context, svc *runtime.EngineVersionService, cfg config.DevSandboxConfig, logger *zap.Logger) error {
-	_, err := svc.Register(ctx, runtime.RegisterInput{
-		Version:  cfg.EngineVersion,
-		ImageRef: cfg.EngineImage,
-	})
-	switch {
-	case err == nil:
-		logger.Info("engine version registered",
-			zap.String("version", cfg.EngineVersion),
-			zap.String("image", cfg.EngineImage),
-		)
-		return nil
-	case errors.Is(err, runtime.ErrEngineVersionTaken):
-		logger.Debug("engine version already registered",
-			zap.String("version", cfg.EngineVersion),
-		)
-		return nil
-	default:
-		return fmt.Errorf("dev_sandbox: register engine version: %w", err)
-	}
-}
-
-// terminalSandboxStatus reports whether a sandbox game has reached a
-// state from which it can no longer be driven back to running. We
-// treat such games as "absent" so the next bootstrap creates a fresh
-// one rather than handing the developer a dead lobby tile.
-func terminalSandboxStatus(status string) bool {
-	switch status {
-	case lobby.GameStatusCancelled, lobby.GameStatusFinished, lobby.GameStatusStartFailed:
-		return true
-	}
-	return false
-}
-
-// purgeTerminalSandboxGames deletes every previous "Dev Sandbox" game
-// the dev user owns that has reached a terminal state
-// (cancelled / finished / start_failed). The cascade declared in
-// `00001_init.sql` removes the matching memberships, applications,
-// invites, runtime records, and player mappings in the same write,
-// so the developer's lobby never piles up dead tiles between
-// `make rebuild` cycles. Non-terminal games are left untouched —
-// a `running` sandbox from a previous boot is the happy path.
-func purgeTerminalSandboxGames(ctx context.Context, svc *lobby.Service, ownerID uuid.UUID, logger *zap.Logger) error {
-	games, err := svc.ListMyGames(ctx, ownerID)
-	if err != nil {
-		return fmt.Errorf("dev_sandbox: list my games: %w", err)
-	}
-	for _, g := range games {
-		if g.GameName != SandboxGameName || g.OwnerUserID == nil || *g.OwnerUserID != ownerID {
-			continue
-		}
-		if !terminalSandboxStatus(g.Status) {
-			continue
-		}
-		if err := svc.DeleteGame(ctx, g.GameID); err != nil {
-			return fmt.Errorf("dev_sandbox: delete terminal sandbox %s: %w", g.GameID, err)
-		}
-		logger.Info("purged terminal sandbox game",
-			zap.String("game_id", g.GameID.String()),
-			zap.String("status", g.Status),
-		)
-	}
-	return nil
-}
-
-func findOrCreateSandboxGame(ctx context.Context, svc *lobby.Service, ownerID uuid.UUID, cfg config.DevSandboxConfig) (lobby.GameRecord, error) {
-	games, err := svc.ListMyGames(ctx, ownerID)
-	if err != nil {
-		return lobby.GameRecord{}, fmt.Errorf("dev_sandbox: list my games: %w", err)
-	}
-	for _, g := range games {
-		if g.GameName != SandboxGameName || g.OwnerUserID == nil || *g.OwnerUserID != ownerID {
-			continue
-		}
-		// `purgeTerminalSandboxGames` ran before us, so any sandbox
-		// game still in the list is either a live one we should
-		// reuse or a transient state we can drive forward.
-		return g, nil
-	}
-	rec, err := svc.CreateGame(ctx, lobby.CreateGameInput{
-		OwnerUserID:         &ownerID,
-		Visibility:          lobby.VisibilityPrivate,
-		GameName:            SandboxGameName,
-		Description:         "Auto-provisioned by backend/internal/devsandbox for solo UI development.",
-		MinPlayers:          int32(cfg.PlayerCount),
-		MaxPlayers:          int32(cfg.PlayerCount),
-		StartGapHours:       0,
-		StartGapPlayers:     0,
-		EnrollmentEndsAt:    time.Now().Add(365 * 24 * time.Hour),
-		TurnSchedule:        SandboxTurnSchedule,
-		TargetEngineVersion: cfg.EngineVersion,
-	})
-	if err != nil {
-		return lobby.GameRecord{}, fmt.Errorf("dev_sandbox: create game: %w", err)
-	}
-	return rec, nil
-}
-
-func ensureMembershipsAndDrive(ctx context.Context, svc *lobby.Service, game lobby.GameRecord, realID uuid.UUID, dummyIDs []uuid.UUID, logger *zap.Logger) (lobby.GameRecord, error) {
-	caller := realID
-	if game.Status == lobby.GameStatusDraft {
-		next, err := svc.OpenEnrollment(ctx, &caller, false, game.GameID)
-		if err != nil {
-			return game, fmt.Errorf("dev_sandbox: open enrollment: %w", err)
-		}
-		game = next
-	}
-
-	if game.Status == lobby.GameStatusEnrollmentOpen {
-		users := append([]uuid.UUID{realID}, dummyIDs...)
-		for i, uid := range users {
-			raceName := fmt.Sprintf("Sandbox-%02d", i+1)
-			if _, err := svc.InsertMembershipDirect(ctx, lobby.InsertMembershipDirectInput{
-				GameID:   game.GameID,
-				UserID:   uid,
-				RaceName: raceName,
-			}); err != nil {
-				return game, fmt.Errorf("dev_sandbox: insert membership %d: %w", i+1, err)
-			}
-		}
-		logger.Info("memberships ensured",
-			zap.Int("count", len(users)),
-			zap.String("game_id", game.GameID.String()),
-		)
-		next, err := svc.ReadyToStart(ctx, &caller, false, game.GameID)
-		if err != nil {
-			return game, fmt.Errorf("dev_sandbox: ready to start: %w", err)
-		}
-		game = next
-	}
-
-	if game.Status == lobby.GameStatusReadyToStart {
-		next, err := svc.Start(ctx, &caller, false, game.GameID)
-		if err != nil {
-			return game, fmt.Errorf("dev_sandbox: start: %w", err)
-		}
-		game = next
-	}
-
-	if game.Status == lobby.GameStatusStartFailed {
-		next, err := svc.RetryStart(ctx, &caller, false, game.GameID)
-		if err != nil {
-			logger.Warn("retry start failed", zap.Error(err))
-			return game, nil
-		}
-		game = next
-		if game.Status == lobby.GameStatusReadyToStart {
-			next, err := svc.Start(ctx, &caller, false, game.GameID)
-			if err != nil {
-				return game, fmt.Errorf("dev_sandbox: start after retry: %w", err)
-			}
-			game = next
-		}
-	}
-
-	return game, nil
-}

backend/internal/devsandbox/bootstrap_test.go

@@ -1,106 +0,0 @@
-package devsandbox
-
-import (
-	"context"
-	"errors"
-	"testing"
-
-	"galaxy/backend/internal/config"
-
-	"github.com/google/uuid"
-	"go.uber.org/zap"
-)
-
-// TestBootstrapSkippedWhenEmailEmpty exercises the no-op branch: with
-// the production posture (Email == "") Bootstrap must return without
-// touching any dependency. The fact that Users/Lobby/EngineVersions
-// are nil here doubles as a check that the early-return runs first.
-func TestBootstrapSkippedWhenEmailEmpty(t *testing.T) {
-	err := Bootstrap(
-		context.Background(),
-		Deps{},
-		config.DevSandboxConfig{},
-		zap.NewNop(),
-	)
-	if err != nil {
-		t.Fatalf("expected nil error on empty email, got: %v", err)
-	}
-}
-
-// TestBootstrapRejectsZeroPlayerCount confirms the validation
-// short-circuits the flow before any DB call when PlayerCount is
-// non-positive but Email is set. The error path is fast and never
-// dereferences the (still-nil) Users/Lobby deps.
-func TestBootstrapRejectsZeroPlayerCount(t *testing.T) {
-	err := Bootstrap(
-		context.Background(),
-		Deps{Users: stubEnsurer{}, Lobby: nil, EngineVersions: nil},
-		config.DevSandboxConfig{
-			Email:         "dev@local.test",
-			EngineImage:   "galaxy-engine:local-dev",
-			EngineVersion: "0.0.0-local-dev",
-			PlayerCount:   0,
-		},
-		zap.NewNop(),
-	)
-	if err == nil {
-		t.Fatal("expected error on zero PlayerCount, got nil")
-	}
-}
-
-// TestBootstrapRejectsMissingDeps checks that a misconfigured wiring
-// (Email set but one of the required services nil) fails fast rather
-// than panicking when the bootstrap reaches its first service call.
-func TestBootstrapRejectsMissingDeps(t *testing.T) {
-	err := Bootstrap(
-		context.Background(),
-		Deps{Users: stubEnsurer{}, Lobby: nil, EngineVersions: nil},
-		config.DevSandboxConfig{
-			Email:         "dev@local.test",
-			EngineImage:   "galaxy-engine:local-dev",
-			EngineVersion: "0.0.0-local-dev",
-			PlayerCount:   20,
-		},
-		zap.NewNop(),
-	)
-	if err == nil {
-		t.Fatal("expected error on missing deps, got nil")
-	}
-	if !errors.Is(err, errMissingDepsSentinel) && err.Error() == "" {
-		// The exact wording is not part of the contract; this branch
-		// only asserts the error is non-nil and human-readable.
-		t.Fatalf("error has empty message: %v", err)
-	}
-}
-
-// errMissingDepsSentinel exists so the assertion above can compile;
-// the real error is constructed via errors.New inside Bootstrap and
-// is intentionally not exported. The test only needs to confirm the
-// returned error has a message.
-var errMissingDepsSentinel = errors.New("sentinel")
-
-// TestTerminalSandboxStatus pins the contract that decides whether a
-// previously created sandbox game gets purged on the next boot.
-// Terminal states are deleted (cascade-style) so the developer's
-// lobby never piles up dead tiles between `make rebuild` cycles.
-func TestTerminalSandboxStatus(t *testing.T) {
-	terminal := []string{"cancelled", "finished", "start_failed"}
-	live := []string{"draft", "enrollment_open", "ready_to_start", "starting", "running", "paused"}
-
-	for _, status := range terminal {
-		if !terminalSandboxStatus(status) {
-			t.Errorf("expected %q to be terminal", status)
-		}
-	}
-	for _, status := range live {
-		if terminalSandboxStatus(status) {
-			t.Errorf("expected %q to be non-terminal", status)
-		}
-	}
-}
-
-type stubEnsurer struct{}
-
-func (stubEnsurer) EnsureByEmail(_ context.Context, _, _, _, _ string) (uuid.UUID, error) {
-	return uuid.UUID{}, nil
-}

backend/internal/lobby/games.go

@@ -274,11 +274,10 @@ func (s *Service) ListFinishedGamesBefore(ctx context.Context, cutoff time.Time)
 // `ON DELETE CASCADE` constraints declared in `00001_init.sql`.
 // Idempotent: returns nil when no game matches.
 //
-// Phase 14 introduces this method for the dev-sandbox bootstrap so a
+// `DeleteGame` is destructive — a hard delete that bypasses the
-// terminal "Dev Sandbox" tile from a previous local-dev session can
+// cascade-notification machinery — so production callers stay on the
-// be scrubbed before a fresh game spawns. Production callers must
+// regular cancel / finish lifecycle. It is exercised by the lobby
-// stay on the regular cancel / finish lifecycle — `DeleteGame` is
+// integration tests.
-// destructive and bypasses the cascade-notification machinery.
 func (s *Service) DeleteGame(ctx context.Context, gameID uuid.UUID) error {
 	if err := s.deps.Store.DeleteGame(ctx, gameID); err != nil {
 		return err

backend/internal/lobby/lobby_e2e_test.go

@@ -248,8 +248,8 @@ func TestEndToEndPrivateGameFlow(t *testing.T) {
 	}
 }
 
-// TestDeleteGameCascadesEverything pins the contract the dev-sandbox
+// TestDeleteGameCascadesEverything pins the DeleteGame contract:
-// bootstrap relies on: removing a game wipes every referencing row
+// removing a game wipes every referencing row
 // (memberships, applications, invites, runtime_records,
 // player_mappings) in a single SQL statement. Before this is wired
 // the developer's lobby pile up cancelled tiles between

backend/internal/lobby/membership_direct.go

@@ -20,9 +20,9 @@ type InsertMembershipDirectInput struct {
 // writes as ApproveApplication: the per-game race-name reservation
 // row plus the membership row, and refreshes the in-memory caches.
 //
-// The method is intended for boot-time provisioning by
+// The method is intended for trusted boot-time provisioning and
-// `backend/internal/devsandbox` and similar trusted callers. It is
+// integration tests; it is not exposed through any HTTP handler. The
-// not exposed through any HTTP handler. The caller must guarantee
+// caller must guarantee
 // game.Status == GameStatusEnrollmentOpen — the function returns
 // ErrConflict otherwise — and that the race-name policy and
 // canonical-key invariants are honoured (the implementation reuses
@@ -30,9 +30,8 @@ type InsertMembershipDirectInput struct {
 // or unsuitable name still fails).
 //
 // Idempotency: if a membership for (GameID, UserID) already exists
-// the function returns the existing row without modifying state.
+// the function returns the existing row without modifying state, so
-// This makes the helper safe to call on every backend boot from
+// the helper is safe to call repeatedly.
-// devsandbox.Bootstrap.
 func (s *Service) InsertMembershipDirect(ctx context.Context, in InsertMembershipDirectInput) (Membership, error) {
 	displayName, err := ValidateDisplayName(in.RaceName)
 	if err != nil {

backend/internal/lobby/store.go

@@ -236,9 +236,8 @@ func (s *Store) ListMyGames(ctx context.Context, userID uuid.UUID) ([]GameRecord
 // referencing table (memberships / applications / invites /
 // runtime_records / player_mappings — all declared with ON DELETE
 // CASCADE in `00001_init.sql`). Idempotent: returns nil when no row
-// matches. Used by the dev-sandbox bootstrap to scrub terminal
+// matches. A hard delete for trusted callers and integration tests;
-// games on every backend boot so the developer's lobby never piles
+// production lifecycle uses cancel / finish.
-// up cancelled tiles.
 func (s *Store) DeleteGame(ctx context.Context, gameID uuid.UUID) error {
 	g := table.Games
 	stmt := g.DELETE().WHERE(g.GameID.EQ(postgres.UUID(gameID)))

backend/internal/opsstatus/opsstatus.go

@@ -0,0 +1,139 @@
+// Package opsstatus reads point-in-time operational signals from Postgres for
+// the admin console dashboard: database reachability, per-status counts of game
+// runtimes, mail deliveries, and notification routes, plus the malformed
+// notification-intent count.
+//
+// It is a read-only projection built entirely through the go-jet query builder
+// against the generated table bindings; it owns no business logic and mutates
+// nothing. Richer, historical metrics are out of scope — those belong to the
+// Prometheus exporters wired on `backend` and `gateway`.
+package opsstatus
+
+import (
+	"context"
+	"database/sql"
+	"fmt"
+	"time"
+
+	"galaxy/backend/internal/postgres/jet/backend/table"
+
+	"github.com/go-jet/jet/v2/postgres"
+)
+
+// defaultCollectTimeout bounds a single Collect call so a slow or wedged
+// database cannot hang the dashboard request.
+const defaultCollectTimeout = 3 * time.Second
+
+// StatusCount pairs a status value with the number of rows currently in it.
+type StatusCount struct {
+	Status string
+	Count  int64
+}
+
+// Snapshot is a point-in-time view of the operational signals rendered on the
+// dashboard. Errors collects per-query failures so a single failing probe
+// degrades to a visible note rather than failing the whole page.
+type Snapshot struct {
+	PostgresHealthy       bool
+	Runtimes              []StatusCount
+	MailDeliveries        []StatusCount
+	NotificationRoutes    []StatusCount
+	NotificationMalformed int64
+	Errors                []string
+}
+
+// Reader collects an operational Snapshot. The admin console depends on this
+// interface so the dashboard can be tested without a database.
+type Reader interface {
+	Collect(ctx context.Context) Snapshot
+}
+
+// Store is the Postgres-backed Reader.
+type Store struct {
+	db      *sql.DB
+	timeout time.Duration
+}
+
+// NewStore constructs a Store reading from db.
+func NewStore(db *sql.DB) *Store {
+	return &Store{db: db, timeout: defaultCollectTimeout}
+}
+
+// Collect gathers the dashboard signals within a bounded timeout. It never
+// returns an error: a failed probe is recorded in Snapshot.Errors and the
+// remaining probes still run, except that a failed Postgres ping short-circuits
+// the rest (the dependent queries would only fail the same way).
+func (s *Store) Collect(ctx context.Context) Snapshot {
+	ctx, cancel := context.WithTimeout(ctx, s.timeout)
+	defer cancel()
+
+	var snap Snapshot
+
+	if err := s.db.PingContext(ctx); err != nil {
+		snap.Errors = append(snap.Errors, fmt.Sprintf("postgres ping: %v", err))
+		return snap
+	}
+	snap.PostgresHealthy = true
+
+	if counts, err := s.statusCounts(ctx, table.RuntimeRecords.Status, table.RuntimeRecords); err != nil {
+		snap.Errors = append(snap.Errors, fmt.Sprintf("runtime status counts: %v", err))
+	} else {
+		snap.Runtimes = counts
+	}
+
+	if counts, err := s.statusCounts(ctx, table.MailDeliveries.Status, table.MailDeliveries); err != nil {
+		snap.Errors = append(snap.Errors, fmt.Sprintf("mail delivery counts: %v", err))
+	} else {
+		snap.MailDeliveries = counts
+	}
+
+	if counts, err := s.statusCounts(ctx, table.NotificationRoutes.Status, table.NotificationRoutes); err != nil {
+		snap.Errors = append(snap.Errors, fmt.Sprintf("notification route counts: %v", err))
+	} else {
+		snap.NotificationRoutes = counts
+	}
+
+	if n, err := s.countAll(ctx, table.NotificationMalformedIntents); err != nil {
+		snap.Errors = append(snap.Errors, fmt.Sprintf("malformed notification count: %v", err))
+	} else {
+		snap.NotificationMalformed = n
+	}
+
+	return snap
+}
+
+// statusCounts runs `SELECT status, COUNT(*) FROM <from> GROUP BY status`
+// through jet and returns the rows ordered by status.
+func (s *Store) statusCounts(ctx context.Context, status postgres.ColumnString, from postgres.ReadableTable) ([]StatusCount, error) {
+	stmt := postgres.SELECT(
+		status.AS("status_count.status"),
+		postgres.COUNT(postgres.STAR).AS("status_count.count"),
+	).FROM(from).GROUP_BY(status).ORDER_BY(status.ASC())
+
+	var rows []struct {
+		Status string `alias:"status_count.status"`
+		Count  int64  `alias:"status_count.count"`
+	}
+	if err := stmt.QueryContext(ctx, s.db, &rows); err != nil {
+		return nil, err
+	}
+
+	out := make([]StatusCount, len(rows))
+	for i, row := range rows {
+		out[i] = StatusCount{Status: row.Status, Count: row.Count}
+	}
+	return out, nil
+}
+
+// countAll runs `SELECT COUNT(*) FROM <from>` through jet.
+func (s *Store) countAll(ctx context.Context, from postgres.ReadableTable) (int64, error) {
+	stmt := postgres.SELECT(postgres.COUNT(postgres.STAR).AS("count")).FROM(from)
+
+	var row struct {
+		Count int64 `alias:"count"`
+	}
+	if err := stmt.QueryContext(ctx, s.db, &row); err != nil {
+		return 0, err
+	}
+	return row.Count, nil
+}

backend/internal/opsstatus/opsstatus_test.go

@@ -0,0 +1,155 @@
+package opsstatus_test
+
+import (
+	"context"
+	"database/sql"
+	"net/url"
+	"testing"
+	"time"
+
+	"galaxy/backend/internal/mail"
+	"galaxy/backend/internal/opsstatus"
+	backendpg "galaxy/backend/internal/postgres"
+	pgshared "galaxy/postgres"
+
+	"github.com/google/uuid"
+	testcontainers "github.com/testcontainers/testcontainers-go"
+	tcpostgres "github.com/testcontainers/testcontainers-go/modules/postgres"
+	"github.com/testcontainers/testcontainers-go/wait"
+)
+
+const (
+	pgImage    = "postgres:16-alpine"
+	pgUser     = "galaxy"
+	pgPassword = "galaxy"
+	pgDatabase = "galaxy_backend"
+	pgSchema   = "backend"
+	pgStartup  = 90 * time.Second
+	pgOpTO     = 10 * time.Second
+)
+
+// startPostgres mirrors the per-package scaffolding used by the other store
+// tests: spin up Postgres, apply migrations, return *sql.DB.
+func startPostgres(t *testing.T) *sql.DB {
+	t.Helper()
+	ctx, cancel := context.WithTimeout(context.Background(), 3*time.Minute)
+	t.Cleanup(cancel)
+
+	pgContainer, err := tcpostgres.Run(ctx, pgImage,
+		tcpostgres.WithDatabase(pgDatabase),
+		tcpostgres.WithUsername(pgUser),
+		tcpostgres.WithPassword(pgPassword),
+		testcontainers.WithWaitStrategy(
+			wait.ForLog("database system is ready to accept connections").
+				WithOccurrence(2).
+				WithStartupTimeout(pgStartup),
+		),
+	)
+	if err != nil {
+		t.Skipf("postgres testcontainer unavailable, skipping: %v", err)
+	}
+	t.Cleanup(func() {
+		if termErr := testcontainers.TerminateContainer(pgContainer); termErr != nil {
+			t.Errorf("terminate postgres container: %v", termErr)
+		}
+	})
+
+	baseDSN, err := pgContainer.ConnectionString(ctx, "sslmode=disable")
+	if err != nil {
+		t.Fatalf("connection string: %v", err)
+	}
+	scopedDSN, err := dsnWithSearchPath(baseDSN, pgSchema)
+	if err != nil {
+		t.Fatalf("scope dsn: %v", err)
+	}
+
+	cfg := pgshared.DefaultConfig()
+	cfg.PrimaryDSN = scopedDSN
+	cfg.OperationTimeout = pgOpTO
+
+	db, err := pgshared.OpenPrimary(ctx, cfg, backendpg.NoObservabilityOptions()...)
+	if err != nil {
+		t.Fatalf("open primary: %v", err)
+	}
+	t.Cleanup(func() { _ = db.Close() })
+
+	if err := pgshared.Ping(ctx, db, cfg.OperationTimeout); err != nil {
+		t.Fatalf("ping: %v", err)
+	}
+	if err := backendpg.ApplyMigrations(ctx, db); err != nil {
+		t.Fatalf("apply migrations: %v", err)
+	}
+	return db
+}
+
+func dsnWithSearchPath(baseDSN, schema string) (string, error) {
+	parsed, err := url.Parse(baseDSN)
+	if err != nil {
+		return "", err
+	}
+	values := parsed.Query()
+	values.Set("search_path", schema)
+	if values.Get("sslmode") == "" {
+		values.Set("sslmode", "disable")
+	}
+	parsed.RawQuery = values.Encode()
+	return parsed.String(), nil
+}
+
+func TestStoreCollect(t *testing.T) {
+	db := startPostgres(t)
+	store := opsstatus.NewStore(db)
+	ctx := context.Background()
+
+	// Empty schema: queries must execute cleanly with zero counts.
+	empty := store.Collect(ctx)
+	if !empty.PostgresHealthy {
+		t.Fatal("PostgresHealthy must be true against a reachable database")
+	}
+	if len(empty.Errors) != 0 {
+		t.Fatalf("unexpected collection errors: %v", empty.Errors)
+	}
+	if got := totalCount(empty.MailDeliveries); got != 0 {
+		t.Fatalf("mail deliveries total = %d, want 0", got)
+	}
+	if len(empty.Runtimes) != 0 || len(empty.NotificationRoutes) != 0 {
+		t.Fatalf("expected empty status slices, got runtimes=%v routes=%v", empty.Runtimes, empty.NotificationRoutes)
+	}
+	if empty.NotificationMalformed != 0 {
+		t.Fatalf("malformed notifications = %d, want 0", empty.NotificationMalformed)
+	}
+
+	// Enqueue one mail delivery and confirm the GROUP BY count reflects it.
+	mailStore := mail.NewStore(db)
+	inserted, err := mailStore.InsertEnqueue(ctx, mail.EnqueueArgs{
+		DeliveryID:     uuid.New(),
+		TemplateID:     mail.TemplateLoginCode,
+		IdempotencyKey: uuid.NewString(),
+		Recipients:     []string{"ops@example.test"},
+		ContentType:    "text/plain",
+		Subject:        "hello",
+		Body:           []byte("hi"),
+	})
+	if err != nil {
+		t.Fatalf("insert mail delivery: %v", err)
+	}
+	if !inserted {
+		t.Fatal("expected the delivery to be inserted")
+	}
+
+	after := store.Collect(ctx)
+	if len(after.Errors) != 0 {
+		t.Fatalf("unexpected collection errors after insert: %v", after.Errors)
+	}
+	if got := totalCount(after.MailDeliveries); got != 1 {
+		t.Fatalf("mail deliveries total after insert = %d, want 1 (statuses: %v)", got, after.MailDeliveries)
+	}
+}
+
+func totalCount(counts []opsstatus.StatusCount) int64 {
+	var total int64
+	for _, c := range counts {
+		total += c.Count
+	}
+	return total
+}

backend/internal/server/handlers_admin_console.go

@@ -0,0 +1,235 @@
+package server
+
+import (
+	"bytes"
+	"net/http"
+	"net/url"
+	"strings"
+
+	"galaxy/backend/internal/adminconsole"
+	"galaxy/backend/internal/opsstatus"
+	"galaxy/backend/internal/server/httperr"
+	"galaxy/backend/internal/server/middleware/basicauth"
+
+	"github.com/gin-gonic/gin"
+	"go.uber.org/zap"
+)
+
+// AdminConsoleHandlers renders the server-side operator console mounted under
+// the `/_gm` route group. It wraps the framework-agnostic
+// adminconsole.Renderer and CSRF signer with the gin glue: the per-page
+// handlers, the embedded static-asset handler, and the CSRF guard middleware
+// applied to state-changing requests. Authentication is provided by the shared
+// admin Basic Auth middleware mounted on the group, so this type assumes the
+// caller has already been verified.
+type AdminConsoleHandlers struct {
+	renderer *adminconsole.Renderer
+	csrf     *adminconsole.CSRF
+	assets   http.Handler
+	monitor        opsstatus.Reader
+	ready          func() bool
+	users          UserAdmin
+	games          GameAdmin
+	runtime        RuntimeAdmin
+	engineVersions EngineVersionAdmin
+	operators      OperatorAdmin
+	mail           MailAdmin
+	notifications  NotificationAdmin
+	diplomail      DiplomailAdmin
+	logger         *zap.Logger
+}
+
+// AdminConsoleDeps bundles the collaborators for the operator console. Every
+// field is optional: a nil Renderer or CSRF falls back to the embedded default
+// templates and a per-process random key; a nil Monitor renders the dashboard
+// without the monitoring panels; a nil Ready reports backend readiness as not
+// ready; a nil Logger falls back to zap.NewNop.
+type AdminConsoleDeps struct {
+	Renderer *adminconsole.Renderer
+	CSRF     *adminconsole.CSRF
+	Monitor        opsstatus.Reader
+	Ready          func() bool
+	Users          UserAdmin
+	Games          GameAdmin
+	Runtime        RuntimeAdmin
+	EngineVersions EngineVersionAdmin
+	Operators      OperatorAdmin
+	Mail           MailAdmin
+	Notifications  NotificationAdmin
+	Diplomail      DiplomailAdmin
+	Logger         *zap.Logger
+}
+
+// NewAdminConsoleHandlers constructs the console handler set from deps. It
+// panics only on conditions that are unrecoverable at startup (template parse
+// failure or crypto/rand failure), both of which indicate a broken build or
+// host rather than a runtime input.
+func NewAdminConsoleHandlers(deps AdminConsoleDeps) *AdminConsoleHandlers {
+	logger := deps.Logger
+	if logger == nil {
+		logger = zap.NewNop()
+	}
+	renderer := deps.Renderer
+	if renderer == nil {
+		renderer = adminconsole.MustNewRenderer()
+	}
+	csrf := deps.CSRF
+	if csrf == nil {
+		generated, err := adminconsole.NewRandomCSRF()
+		if err != nil {
+			panic(err)
+		}
+		csrf = generated
+	}
+
+	assetsFS, err := adminconsole.Assets()
+	if err != nil {
+		panic(err)
+	}
+
+	return &AdminConsoleHandlers{
+		renderer: renderer,
+		csrf:     csrf,
+		assets:   http.StripPrefix("/_gm/assets/", http.FileServer(http.FS(assetsFS))),
+		monitor:        deps.Monitor,
+		ready:          deps.Ready,
+		users:          deps.Users,
+		games:          deps.Games,
+		runtime:        deps.Runtime,
+		engineVersions: deps.EngineVersions,
+		operators:      deps.Operators,
+		mail:           deps.Mail,
+		notifications:  deps.Notifications,
+		diplomail:      deps.Diplomail,
+		logger:         logger.Named("http.admin.console"),
+	}
+}
+
+// Dashboard renders the console landing page (GET /_gm and GET /_gm/),
+// including the monitoring panels when an ops-status reader is wired.
+func (h *AdminConsoleHandlers) Dashboard() gin.HandlerFunc {
+	return func(c *gin.Context) {
+		data := adminconsole.DashboardData{}
+		if h.ready != nil {
+			data.BackendReady = h.ready()
+		}
+		if h.monitor != nil {
+			data.MonitorAvailable = true
+			snapshot := h.monitor.Collect(c.Request.Context())
+			data.PostgresHealthy = snapshot.PostgresHealthy
+			data.Runtimes = toViewCounts(snapshot.Runtimes)
+			data.MailDeliveries = toViewCounts(snapshot.MailDeliveries)
+			data.NotificationRoutes = toViewCounts(snapshot.NotificationRoutes)
+			data.NotificationMalformed = snapshot.NotificationMalformed
+			data.Errors = snapshot.Errors
+		}
+		h.render(c, http.StatusOK, "dashboard", "dashboard", "Dashboard", data)
+	}
+}
+
+// toViewCounts maps ops-status counts to the console's view-layer counts.
+func toViewCounts(in []opsstatus.StatusCount) []adminconsole.StatusCount {
+	if len(in) == 0 {
+		return nil
+	}
+	out := make([]adminconsole.StatusCount, len(in))
+	for i, sc := range in {
+		out[i] = adminconsole.StatusCount{Status: sc.Status, Count: sc.Count}
+	}
+	return out
+}
+
+// Asset serves the embedded console static assets under `/_gm/assets/`.
+func (h *AdminConsoleHandlers) Asset() gin.HandlerFunc {
+	return gin.WrapH(h.assets)
+}
+
+// RequireCSRF returns middleware guarding state-changing requests against
+// cross-site request forgery. Safe methods pass through untouched. For unsafe
+// methods it requires both a same-origin Origin/Referer header (when the
+// browser sends one) and a valid per-operator token in the `_csrf` form field;
+// either check failing yields 403.
+func (h *AdminConsoleHandlers) RequireCSRF() gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if isSafeHTTPMethod(c.Request.Method) {
+			c.Next()
+			return
+		}
+		if !sameOriginRequest(c.Request) {
+			httperr.Abort(c, http.StatusForbidden, httperr.CodeForbidden, "cross-origin request rejected")
+			return
+		}
+		username, _ := basicauth.UsernameFromContext(c.Request.Context())
+		if !h.csrf.Verify(username, c.PostForm("_csrf")) {
+			httperr.Abort(c, http.StatusForbidden, httperr.CodeForbidden, "invalid or missing CSRF token")
+			return
+		}
+		c.Next()
+	}
+}
+
+// render composes the data common to every console page (operator name, CSRF
+// token, active navigation entry) and writes the named page. It renders into an
+// intermediate buffer so a template failure surfaces as a clean 500 without
+// emitting a partial document.
+func (h *AdminConsoleHandlers) render(c *gin.Context, status int, page, activeNav, title string, data any) {
+	username, _ := basicauth.UsernameFromContext(c.Request.Context())
+
+	var buf bytes.Buffer
+	err := h.renderer.Render(&buf, page, adminconsole.PageData{
+		Title:     title,
+		Username:  username,
+		CSRFToken: h.csrf.Token(username),
+		ActiveNav: activeNav,
+		Data:      data,
+	})
+	if err != nil {
+		h.logger.Error("render admin console page", zap.String("page", page), zap.Error(err))
+		httperr.Abort(c, http.StatusInternalServerError, httperr.CodeInternalError, "failed to render page")
+		return
+	}
+
+	c.Data(status, "text/html; charset=utf-8", buf.Bytes())
+}
+
+// renderMessage renders the generic message page (not-found, validation, or
+// operation-failure notices). class selects the CSS styling and backHref, when
+// non-empty, adds a back link.
+func (h *AdminConsoleHandlers) renderMessage(c *gin.Context, status int, activeNav, title, message, class, backHref string) {
+	h.render(c, status, "message", activeNav, title, adminconsole.MessageData{
+		Message:  message,
+		Class:    class,
+		BackHref: backHref,
+	})
+}
+
+// isSafeHTTPMethod reports whether method is a read-only HTTP method that the
+// CSRF guard may let through without a token.
+func isSafeHTTPMethod(method string) bool {
+	switch method {
+	case http.MethodGet, http.MethodHead, http.MethodOptions:
+		return true
+	default:
+		return false
+	}
+}
+
+// sameOriginRequest reports whether the request's Origin (or, failing that,
+// Referer) names the same host as the request itself. A request that carries
+// neither header is treated as same-origin, leaving the CSRF token as the sole
+// guard; a malformed or cross-host value is rejected. This relies on the
+// gateway reverse proxy preserving the inbound Host header.
+func sameOriginRequest(r *http.Request) bool {
+	source := r.Header.Get("Origin")
+	if source == "" {
+		source = r.Header.Get("Referer")
+	}
+	if source == "" {
+		return true
+	}
+	parsed, err := url.Parse(source)
+	if err != nil || parsed.Host == "" {
+		return false
+	}
+	return strings.EqualFold(parsed.Host, r.Host)
+}

backend/internal/server/handlers_admin_console_games.go

@@ -0,0 +1,423 @@
+package server
+
+import (
+	"context"
+	"errors"
+	"net/http"
+	"strconv"
+	"strings"
+	"time"
+
+	"galaxy/backend/internal/adminconsole"
+	"galaxy/backend/internal/lobby"
+	"galaxy/backend/internal/runtime"
+
+	"github.com/gin-gonic/gin"
+	"github.com/google/uuid"
+	"go.uber.org/zap"
+)
+
+// GameAdmin is the subset of the lobby service the console uses for games.
+type GameAdmin interface {
+	ListAdminGames(ctx context.Context, page, pageSize int) (lobby.GamePage, error)
+	GetGame(ctx context.Context, gameID uuid.UUID) (lobby.GameRecord, error)
+	CreateGame(ctx context.Context, input lobby.CreateGameInput) (lobby.GameRecord, error)
+	AdminForceStart(ctx context.Context, gameID uuid.UUID) (lobby.GameRecord, error)
+	AdminForceStop(ctx context.Context, gameID uuid.UUID) (lobby.GameRecord, error)
+	AdminBanMember(ctx context.Context, gameID, userID uuid.UUID, reason string) (lobby.Membership, error)
+}
+
+// RuntimeAdmin is the subset of the runtime service the console uses.
+type RuntimeAdmin interface {
+	GetRuntime(ctx context.Context, gameID uuid.UUID) (runtime.RuntimeRecord, error)
+	AdminRestart(ctx context.Context, gameID uuid.UUID) (runtime.OperationLog, error)
+	AdminPatch(ctx context.Context, gameID uuid.UUID, targetVersion string) (runtime.OperationLog, error)
+	AdminForceNextTurn(ctx context.Context, gameID uuid.UUID) (runtime.OperationLog, error)
+}
+
+// EngineVersionAdmin is the subset of the engine-version service the console uses.
+type EngineVersionAdmin interface {
+	List(ctx context.Context) ([]runtime.EngineVersion, error)
+	Register(ctx context.Context, in runtime.RegisterInput) (runtime.EngineVersion, error)
+	Disable(ctx context.Context, version string) (runtime.EngineVersion, error)
+}
+
+// GamesList renders GET /_gm/games.
+func (h *AdminConsoleHandlers) GamesList() gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if h.games == nil {
+			h.renderMessage(c, http.StatusServiceUnavailable, "games", "Games", "Game administration is not available.", "bad", "/_gm/")
+			return
+		}
+		page := parsePositiveQueryInt(c.Query("page"), 1)
+		pageSize := parsePositiveQueryInt(c.Query("page_size"), 50)
+		result, err := h.games.ListAdminGames(c.Request.Context(), page, pageSize)
+		if err != nil {
+			h.logger.Error("admin console: list games", zap.Error(err))
+			h.renderMessage(c, http.StatusInternalServerError, "games", "Games", "Failed to load games.", "bad", "/_gm/")
+			return
+		}
+		h.render(c, http.StatusOK, "games", "games", "Games", toGamesListData(result))
+	}
+}
+
+// GameCreate handles POST /_gm/games — create a public game.
+func (h *AdminConsoleHandlers) GameCreate() gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if h.games == nil {
+			h.renderMessage(c, http.StatusServiceUnavailable, "games", "Games", "Game administration is not available.", "bad", "/_gm/")
+			return
+		}
+		enrollmentEndsAt, err := parseConsoleDateTime(c.PostForm("enrollment_ends_at"))
+		if err != nil {
+			h.renderMessage(c, http.StatusBadRequest, "games", "Invalid input", "Enrollment end must be a valid date/time.", "bad", "/_gm/games")
+			return
+		}
+		game, err := h.games.CreateGame(c.Request.Context(), lobby.CreateGameInput{
+			OwnerUserID:         nil,
+			Visibility:          lobby.VisibilityPublic,
+			GameName:            strings.TrimSpace(c.PostForm("game_name")),
+			Description:         strings.TrimSpace(c.PostForm("description")),
+			MinPlayers:          formInt32(c, "min_players"),
+			MaxPlayers:          formInt32(c, "max_players"),
+			StartGapHours:       formInt32(c, "start_gap_hours"),
+			StartGapPlayers:     formInt32(c, "start_gap_players"),
+			EnrollmentEndsAt:    enrollmentEndsAt,
+			TurnSchedule:        strings.TrimSpace(c.PostForm("turn_schedule")),
+			TargetEngineVersion: strings.TrimSpace(c.PostForm("target_engine_version")),
+		})
+		if err != nil {
+			if errors.Is(err, lobby.ErrInvalidInput) {
+				h.renderMessage(c, http.StatusBadRequest, "games", "Invalid input", "The game could not be created: check the fields.", "bad", "/_gm/games")
+				return
+			}
+			h.logger.Error("admin console: create game", zap.Error(err))
+			h.renderMessage(c, http.StatusInternalServerError, "games", "Create failed", "Failed to create the game.", "bad", "/_gm/games")
+			return
+		}
+		c.Redirect(http.StatusSeeOther, "/_gm/games/"+game.GameID.String())
+	}
+}
+
+// GameDetail renders GET /_gm/games/:game_id with the runtime snapshot.
+func (h *AdminConsoleHandlers) GameDetail() gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if h.games == nil {
+			h.renderMessage(c, http.StatusServiceUnavailable, "games", "Games", "Game administration is not available.", "bad", "/_gm/")
+			return
+		}
+		gameID, ok := parseGameIDParam(c)
+		if !ok {
+			return
+		}
+		game, err := h.games.GetGame(c.Request.Context(), gameID)
+		if err != nil {
+			if errors.Is(err, lobby.ErrNotFound) {
+				h.renderMessage(c, http.StatusNotFound, "games", "Game not found", "No such game.", "bad", "/_gm/games")
+				return
+			}
+			h.logger.Error("admin console: get game", zap.Error(err))
+			h.renderMessage(c, http.StatusInternalServerError, "games", "Games", "Failed to load the game.", "bad", "/_gm/games")
+			return
+		}
+
+		var runtimeRecord *runtime.RuntimeRecord
+		if h.runtime != nil {
+			if record, rtErr := h.runtime.GetRuntime(c.Request.Context(), gameID); rtErr == nil {
+				runtimeRecord = &record
+			}
+		}
+		h.render(c, http.StatusOK, "game_detail", "games", game.GameName, toGameDetailData(game, runtimeRecord))
+	}
+}
+
+// GameForceStart handles POST /_gm/games/:game_id/force-start.
+func (h *AdminConsoleHandlers) GameForceStart() gin.HandlerFunc {
+	return h.gameAction("force-start", func(ctx context.Context, gameID uuid.UUID) error {
+		_, err := h.games.AdminForceStart(ctx, gameID)
+		return err
+	})
+}
+
+// GameForceStop handles POST /_gm/games/:game_id/force-stop.
+func (h *AdminConsoleHandlers) GameForceStop() gin.HandlerFunc {
+	return h.gameAction("force-stop", func(ctx context.Context, gameID uuid.UUID) error {
+		_, err := h.games.AdminForceStop(ctx, gameID)
+		return err
+	})
+}
+
+// GameBanMember handles POST /_gm/games/:game_id/ban-member.
+func (h *AdminConsoleHandlers) GameBanMember() gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if h.games == nil {
+			h.renderMessage(c, http.StatusServiceUnavailable, "games", "Games", "Game administration is not available.", "bad", "/_gm/")
+			return
+		}
+		gameID, ok := parseGameIDParam(c)
+		if !ok {
+			return
+		}
+		back := "/_gm/games/" + gameID.String()
+		userID, err := uuid.Parse(strings.TrimSpace(c.PostForm("user_id")))
+		if err != nil {
+			h.renderMessage(c, http.StatusBadRequest, "games", "Invalid input", "User ID must be a valid UUID.", "bad", back)
+			return
+		}
+		if _, err := h.games.AdminBanMember(c.Request.Context(), gameID, userID, strings.TrimSpace(c.PostForm("reason"))); err != nil {
+			h.logger.Error("admin console: ban member", zap.Error(err))
+			h.renderMessage(c, http.StatusInternalServerError, "games", "Ban failed", "Failed to ban the member.", "bad", back)
+			return
+		}
+		c.Redirect(http.StatusSeeOther, back)
+	}
+}
+
+// RuntimeRestart handles POST /_gm/games/:game_id/runtime/restart.
+func (h *AdminConsoleHandlers) RuntimeRestart() gin.HandlerFunc {
+	return h.runtimeAction("restart", func(ctx context.Context, gameID uuid.UUID) error {
+		_, err := h.runtime.AdminRestart(ctx, gameID)
+		return err
+	})
+}
+
+// RuntimeForceNextTurn handles POST /_gm/games/:game_id/runtime/force-next-turn.
+func (h *AdminConsoleHandlers) RuntimeForceNextTurn() gin.HandlerFunc {
+	return h.runtimeAction("force-next-turn", func(ctx context.Context, gameID uuid.UUID) error {
+		_, err := h.runtime.AdminForceNextTurn(ctx, gameID)
+		return err
+	})
+}
+
+// RuntimePatch handles POST /_gm/games/:game_id/runtime/patch.
+func (h *AdminConsoleHandlers) RuntimePatch() gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if h.runtime == nil {
+			h.renderMessage(c, http.StatusServiceUnavailable, "games", "Runtime", "Runtime administration is not available.", "bad", "/_gm/games")
+			return
+		}
+		gameID, ok := parseGameIDParam(c)
+		if !ok {
+			return
+		}
+		back := "/_gm/games/" + gameID.String()
+		target := strings.TrimSpace(c.PostForm("target_version"))
+		if target == "" {
+			h.renderMessage(c, http.StatusBadRequest, "games", "Invalid input", "A target version is required.", "bad", back)
+			return
+		}
+		if _, err := h.runtime.AdminPatch(c.Request.Context(), gameID, target); err != nil {
+			h.logger.Error("admin console: runtime patch", zap.Error(err))
+			h.renderMessage(c, http.StatusInternalServerError, "games", "Patch failed", "Failed to patch the runtime.", "bad", back)
+			return
+		}
+		c.Redirect(http.StatusSeeOther, back)
+	}
+}
+
+// gameAction is the shared shape for game-state POST actions that take only the
+// game id and redirect back to the detail page.
+func (h *AdminConsoleHandlers) gameAction(label string, run func(context.Context, uuid.UUID) error) gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if h.games == nil {
+			h.renderMessage(c, http.StatusServiceUnavailable, "games", "Games", "Game administration is not available.", "bad", "/_gm/")
+			return
+		}
+		gameID, ok := parseGameIDParam(c)
+		if !ok {
+			return
+		}
+		back := "/_gm/games/" + gameID.String()
+		if err := run(c.Request.Context(), gameID); err != nil {
+			h.logger.Error("admin console: game "+label, zap.Error(err))
+			h.renderMessage(c, http.StatusInternalServerError, "games", "Action failed", "The "+label+" action failed.", "bad", back)
+			return
+		}
+		c.Redirect(http.StatusSeeOther, back)
+	}
+}
+
+// runtimeAction is the shared shape for runtime POST actions that take only the
+// game id and redirect back to the detail page.
+func (h *AdminConsoleHandlers) runtimeAction(label string, run func(context.Context, uuid.UUID) error) gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if h.runtime == nil {
+			h.renderMessage(c, http.StatusServiceUnavailable, "games", "Runtime", "Runtime administration is not available.", "bad", "/_gm/games")
+			return
+		}
+		gameID, ok := parseGameIDParam(c)
+		if !ok {
+			return
+		}
+		back := "/_gm/games/" + gameID.String()
+		if err := run(c.Request.Context(), gameID); err != nil {
+			h.logger.Error("admin console: runtime "+label, zap.Error(err))
+			h.renderMessage(c, http.StatusInternalServerError, "games", "Action failed", "The runtime "+label+" action failed.", "bad", back)
+			return
+		}
+		c.Redirect(http.StatusSeeOther, back)
+	}
+}
+
+// EngineVersionsList renders GET /_gm/engine-versions.
+func (h *AdminConsoleHandlers) EngineVersionsList() gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if h.engineVersions == nil {
+			h.renderMessage(c, http.StatusServiceUnavailable, "engine-versions", "Engine versions", "Engine-version administration is not available.", "bad", "/_gm/")
+			return
+		}
+		items, err := h.engineVersions.List(c.Request.Context())
+		if err != nil {
+			h.logger.Error("admin console: list engine versions", zap.Error(err))
+			h.renderMessage(c, http.StatusInternalServerError, "engine-versions", "Engine versions", "Failed to load engine versions.", "bad", "/_gm/")
+			return
+		}
+		h.render(c, http.StatusOK, "engine_versions", "games", "Engine versions", toEngineVersionsData(items))
+	}
+}
+
+// EngineVersionRegister handles POST /_gm/engine-versions.
+func (h *AdminConsoleHandlers) EngineVersionRegister() gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if h.engineVersions == nil {
+			h.renderMessage(c, http.StatusServiceUnavailable, "engine-versions", "Engine versions", "Engine-version administration is not available.", "bad", "/_gm/")
+			return
+		}
+		enabled := c.PostForm("enabled") == "true"
+		_, err := h.engineVersions.Register(c.Request.Context(), runtime.RegisterInput{
+			Version:  strings.TrimSpace(c.PostForm("version")),
+			ImageRef: strings.TrimSpace(c.PostForm("image_ref")),
+			Enabled:  &enabled,
+		})
+		if err != nil {
+			if errors.Is(err, runtime.ErrInvalidInput) || errors.Is(err, runtime.ErrEngineVersionTaken) {
+				h.renderMessage(c, http.StatusBadRequest, "engine-versions", "Invalid input", "The version could not be registered (invalid semver, missing image, or duplicate).", "bad", "/_gm/engine-versions")
+				return
+			}
+			h.logger.Error("admin console: register engine version", zap.Error(err))
+			h.renderMessage(c, http.StatusInternalServerError, "engine-versions", "Register failed", "Failed to register the engine version.", "bad", "/_gm/engine-versions")
+			return
+		}
+		c.Redirect(http.StatusSeeOther, "/_gm/engine-versions")
+	}
+}
+
+// EngineVersionDisable handles POST /_gm/engine-versions/:version/disable.
+func (h *AdminConsoleHandlers) EngineVersionDisable() gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if h.engineVersions == nil {
+			h.renderMessage(c, http.StatusServiceUnavailable, "engine-versions", "Engine versions", "Engine-version administration is not available.", "bad", "/_gm/")
+			return
+		}
+		version := strings.TrimSpace(c.Param("version"))
+		if _, err := h.engineVersions.Disable(c.Request.Context(), version); err != nil {
+			h.logger.Error("admin console: disable engine version", zap.Error(err))
+			h.renderMessage(c, http.StatusInternalServerError, "engine-versions", "Disable failed", "Failed to disable the engine version.", "bad", "/_gm/engine-versions")
+			return
+		}
+		c.Redirect(http.StatusSeeOther, "/_gm/engine-versions")
+	}
+}
+
+// formInt32 reads a non-negative int32 form field, defaulting to 0.
+func formInt32(c *gin.Context, name string) int32 {
+	parsed, err := strconv.Atoi(strings.TrimSpace(c.PostForm(name)))
+	if err != nil || parsed < 0 {
+		return 0
+	}
+	return int32(parsed)
+}
+
+// parseConsoleDateTime parses the value of an <input type="datetime-local">
+// (or an RFC 3339 timestamp) as UTC.
+func parseConsoleDateTime(raw string) (time.Time, error) {
+	raw = strings.TrimSpace(raw)
+	for _, layout := range []string{"2006-01-02T15:04", "2006-01-02T15:04:05", time.RFC3339} {
+		if t, err := time.ParseInLocation(layout, raw, time.UTC); err == nil {
+			return t.UTC(), nil
+		}
+	}
+	return time.Time{}, errors.New("invalid date/time")
+}
+
+// toGamesListData maps a game page into the games list view model.
+func toGamesListData(page lobby.GamePage) adminconsole.GamesListData {
+	data := adminconsole.GamesListData{
+		Items:    make([]adminconsole.GameRow, 0, len(page.Items)),
+		Page:     page.Page,
+		PageSize: page.PageSize,
+		Total:    page.Total,
+		PrevPage: page.Page - 1,
+		NextPage: page.Page + 1,
+		HasPrev:  page.Page > 1,
+		HasNext:  page.Page*page.PageSize < page.Total,
+	}
+	for _, game := range page.Items {
+		data.Items = append(data.Items, adminconsole.GameRow{
+			GameID:       game.GameID.String(),
+			GameName:     game.GameName,
+			Visibility:   game.Visibility,
+			Status:       game.Status,
+			Owner:        ownerLabel(game.OwnerUserID),
+			Players:      strconv.Itoa(int(game.MinPlayers)) + "–" + strconv.Itoa(int(game.MaxPlayers)),
+			TurnSchedule: game.TurnSchedule,
+			CreatedAt:    fmtConsoleTime(game.CreatedAt),
+		})
+	}
+	return data
+}
+
+// toGameDetailData maps a game record and optional runtime record into the
+// detail view model.
+func toGameDetailData(game lobby.GameRecord, rec *runtime.RuntimeRecord) adminconsole.GameDetailData {
+	data := adminconsole.GameDetailData{
+		GameID:              game.GameID.String(),
+		GameName:            game.GameName,
+		Description:         game.Description,
+		Visibility:          game.Visibility,
+		Status:              game.Status,
+		Owner:               ownerLabel(game.OwnerUserID),
+		MinPlayers:          game.MinPlayers,
+		MaxPlayers:          game.MaxPlayers,
+		StartGapHours:       game.StartGapHours,
+		StartGapPlayers:     game.StartGapPlayers,
+		TurnSchedule:        game.TurnSchedule,
+		TargetEngineVersion: game.TargetEngineVersion,
+		EnrollmentEndsAt:    fmtConsoleTime(game.EnrollmentEndsAt),
+		CreatedAt:           fmtConsoleTime(game.CreatedAt),
+		StartedAt:           fmtConsoleTimePtr(game.StartedAt),
+		FinishedAt:          fmtConsoleTimePtr(game.FinishedAt),
+	}
+	if rec != nil {
+		data.HasRuntime = true
+		data.RuntimeStatus = rec.Status
+		data.CurrentEngineVersion = rec.CurrentEngineVersion
+		data.EngineHealth = rec.EngineHealth
+		data.CurrentTurn = rec.CurrentTurn
+		data.NextGenerationAt = fmtConsoleTimePtr(rec.NextGenerationAt)
+		data.Paused = rec.Paused
+	}
+	return data
+}
+
+// toEngineVersionsData maps engine versions into the registry view model.
+func toEngineVersionsData(items []runtime.EngineVersion) adminconsole.EngineVersionsData {
+	data := adminconsole.EngineVersionsData{Items: make([]adminconsole.EngineVersionRow, 0, len(items))}
+	for _, v := range items {
+		data.Items = append(data.Items, adminconsole.EngineVersionRow{
+			Version:   v.Version,
+			ImageRef:  v.ImageRef,
+			Enabled:   v.Enabled,
+			CreatedAt: fmtConsoleTime(v.CreatedAt),
+		})
+	}
+	return data
+}
+
+// ownerLabel renders an optional owner id; public games have no owner.
+func ownerLabel(ownerID *uuid.UUID) string {
+	if ownerID == nil {
+		return "—"
+	}
+	return ownerID.String()
+}

backend/internal/server/handlers_admin_console_games_test.go

@@ -0,0 +1,353 @@
+package server
+
+import (
+	"context"
+	"errors"
+	"net/http"
+	"net/http/httptest"
+	"strings"
+	"testing"
+
+	"galaxy/backend/internal/adminconsole"
+	"galaxy/backend/internal/lobby"
+	"galaxy/backend/internal/runtime"
+	"galaxy/backend/internal/server/middleware/basicauth"
+
+	"github.com/google/uuid"
+	"go.uber.org/zap"
+)
+
+type fakeGameAdmin struct {
+	page    lobby.GamePage
+	game    lobby.GameRecord
+	getErr  error
+	created lobby.CreateGameInput
+
+	createCalls     int
+	forceStartCalls int
+	forceStopCalls  int
+	banCalls        int
+	lastBanUser     uuid.UUID
+	lastBanReason   string
+}
+
+func (f *fakeGameAdmin) ListAdminGames(context.Context, int, int) (lobby.GamePage, error) {
+	return f.page, nil
+}
+func (f *fakeGameAdmin) GetGame(context.Context, uuid.UUID) (lobby.GameRecord, error) {
+	return f.game, f.getErr
+}
+func (f *fakeGameAdmin) CreateGame(_ context.Context, in lobby.CreateGameInput) (lobby.GameRecord, error) {
+	f.createCalls++
+	f.created = in
+	return f.game, nil
+}
+func (f *fakeGameAdmin) AdminForceStart(context.Context, uuid.UUID) (lobby.GameRecord, error) {
+	f.forceStartCalls++
+	return f.game, nil
+}
+func (f *fakeGameAdmin) AdminForceStop(context.Context, uuid.UUID) (lobby.GameRecord, error) {
+	f.forceStopCalls++
+	return f.game, nil
+}
+func (f *fakeGameAdmin) AdminBanMember(_ context.Context, _, userID uuid.UUID, reason string) (lobby.Membership, error) {
+	f.banCalls++
+	f.lastBanUser = userID
+	f.lastBanReason = reason
+	return lobby.Membership{}, nil
+}
+
+type fakeRuntimeAdmin struct {
+	record           runtime.RuntimeRecord
+	getErr           error
+	restartCalls     int
+	forceNextCalls   int
+	patchCalls       int
+	lastPatchVersion string
+}
+
+func (f *fakeRuntimeAdmin) GetRuntime(context.Context, uuid.UUID) (runtime.RuntimeRecord, error) {
+	return f.record, f.getErr
+}
+func (f *fakeRuntimeAdmin) AdminRestart(context.Context, uuid.UUID) (runtime.OperationLog, error) {
+	f.restartCalls++
+	return runtime.OperationLog{}, nil
+}
+func (f *fakeRuntimeAdmin) AdminPatch(_ context.Context, _ uuid.UUID, target string) (runtime.OperationLog, error) {
+	f.patchCalls++
+	f.lastPatchVersion = target
+	return runtime.OperationLog{}, nil
+}
+func (f *fakeRuntimeAdmin) AdminForceNextTurn(context.Context, uuid.UUID) (runtime.OperationLog, error) {
+	f.forceNextCalls++
+	return runtime.OperationLog{}, nil
+}
+
+type fakeEngineVersionAdmin struct {
+	list          []runtime.EngineVersion
+	registered    runtime.RegisterInput
+	registerCalls int
+	disableCalls  int
+	lastDisabled  string
+}
+
+func (f *fakeEngineVersionAdmin) List(context.Context) ([]runtime.EngineVersion, error) {
+	return f.list, nil
+}
+func (f *fakeEngineVersionAdmin) Register(_ context.Context, in runtime.RegisterInput) (runtime.EngineVersion, error) {
+	f.registerCalls++
+	f.registered = in
+	return runtime.EngineVersion{}, nil
+}
+func (f *fakeEngineVersionAdmin) Disable(_ context.Context, version string) (runtime.EngineVersion, error) {
+	f.disableCalls++
+	f.lastDisabled = version
+	return runtime.EngineVersion{}, nil
+}
+
+func newGamesConsoleRouter(t *testing.T, games GameAdmin, rt RuntimeAdmin, ev EngineVersionAdmin) (http.Handler, *adminconsole.CSRF) {
+	t.Helper()
+	csrf := adminconsole.NewCSRF([]byte("test-key"))
+	handler, err := NewRouter(RouterDependencies{
+		Logger:        zap.NewNop(),
+		AdminVerifier: basicauth.NewStaticVerifier("secret"),
+		AdminConsole: NewAdminConsoleHandlers(AdminConsoleDeps{
+			CSRF: csrf, Games: games, Runtime: rt, EngineVersions: ev,
+		}),
+	})
+	if err != nil {
+		t.Fatalf("NewRouter: %v", err)
+	}
+	return handler, csrf
+}
+
+func consoleGet(t *testing.T, router http.Handler, path string) *httptest.ResponseRecorder {
+	t.Helper()
+	req := httptest.NewRequest(http.MethodGet, path, nil)
+	req.SetBasicAuth("ops", "secret")
+	rec := httptest.NewRecorder()
+	router.ServeHTTP(rec, req)
+	return rec
+}
+
+func consolePost(t *testing.T, router http.Handler, path, form string) *httptest.ResponseRecorder {
+	t.Helper()
+	req := httptest.NewRequest(http.MethodPost, "http://galaxy.lan"+path, strings.NewReader(form))
+	req.Header.Set("Content-Type", "application/x-www-form-urlencoded")
+	req.Header.Set("Origin", "https://galaxy.lan")
+	req.SetBasicAuth("ops", "secret")
+	rec := httptest.NewRecorder()
+	router.ServeHTTP(rec, req)
+	return rec
+}
+
+func TestConsoleGamesList(t *testing.T) {
+	games := &fakeGameAdmin{page: lobby.GamePage{
+		Items: []lobby.GameRecord{{GameID: uuid.New(), GameName: "Nova", Visibility: "public", Status: "enrollment_open"}},
+		Page:  1, PageSize: 50, Total: 1,
+	}}
+	router, _ := newGamesConsoleRouter(t, games, nil, nil)
+
+	rec := consoleGet(t, router, "/_gm/games")
+	if rec.Code != http.StatusOK {
+		t.Fatalf("status = %d, want 200; body=%s", rec.Code, rec.Body.String())
+	}
+	for _, want := range []string{"Nova", "public", "enrollment_open", "Create public game"} {
+		if !strings.Contains(rec.Body.String(), want) {
+			t.Errorf("games list missing %q", want)
+		}
+	}
+}
+
+func TestConsoleGameDetailWithRuntime(t *testing.T) {
+	id := uuid.New()
+	games := &fakeGameAdmin{game: lobby.GameRecord{GameID: id, GameName: "Nova", Status: "running"}}
+	rt := &fakeRuntimeAdmin{record: runtime.RuntimeRecord{GameID: id, Status: "running", CurrentEngineVersion: "0.1.0", CurrentTurn: 7}}
+	router, csrf := newGamesConsoleRouter(t, games, rt, nil)
+
+	rec := consoleGet(t, router, "/_gm/games/"+id.String())
+	if rec.Code != http.StatusOK {
+		t.Fatalf("status = %d, want 200; body=%s", rec.Code, rec.Body.String())
+	}
+	for _, want := range []string{"Nova", "Force start", "Force stop", "0.1.0", "Patch", "Ban member", csrf.Token("ops")} {
+		if !strings.Contains(rec.Body.String(), want) {
+			t.Errorf("game detail missing %q", want)
+		}
+	}
+}
+
+func TestConsoleGameDetailNoRuntime(t *testing.T) {
+	id := uuid.New()
+	games := &fakeGameAdmin{game: lobby.GameRecord{GameID: id, GameName: "Nova"}}
+	rt := &fakeRuntimeAdmin{getErr: errors.New("not found")}
+	router, _ := newGamesConsoleRouter(t, games, rt, nil)
+
+	rec := consoleGet(t, router, "/_gm/games/"+id.String())
+	if rec.Code != http.StatusOK {
+		t.Fatalf("status = %d, want 200", rec.Code)
+	}
+	if !strings.Contains(rec.Body.String(), "No runtime record") {
+		t.Error("expected a no-runtime note")
+	}
+}
+
+func TestConsoleGameCreate(t *testing.T) {
+	id := uuid.New()
+	games := &fakeGameAdmin{game: lobby.GameRecord{GameID: id}}
+	router, csrf := newGamesConsoleRouter(t, games, nil, nil)
+
+	form := "_csrf=" + csrf.Token("ops") +
+		"&game_name=Nova&description=d&min_players=2&max_players=8&start_gap_hours=0&start_gap_players=0" +
+		"&enrollment_ends_at=2030-01-02T15:04&turn_schedule=@every+24h&target_engine_version=0.1.0"
+	rec := consolePost(t, router, "/_gm/games", form)
+
+	if rec.Code != http.StatusSeeOther {
+		t.Fatalf("status = %d, want 303; body=%s", rec.Code, rec.Body.String())
+	}
+	if got := rec.Header().Get("Location"); got != "/_gm/games/"+id.String() {
+		t.Errorf("redirect = %q, want detail page", got)
+	}
+	if games.createCalls != 1 {
+		t.Fatalf("CreateGame called %d times, want 1", games.createCalls)
+	}
+	if games.created.Visibility != lobby.VisibilityPublic {
+		t.Errorf("visibility = %q, want public", games.created.Visibility)
+	}
+	if games.created.GameName != "Nova" {
+		t.Errorf("game name = %q", games.created.GameName)
+	}
+	if games.created.EnrollmentEndsAt.Year() != 2030 {
+		t.Errorf("enrollment year = %d, want 2030", games.created.EnrollmentEndsAt.Year())
+	}
+	if games.created.OwnerUserID != nil {
+		t.Error("public game must have a nil owner")
+	}
+}
+
+func TestConsoleGameForceStart(t *testing.T) {
+	id := uuid.New()
+	games := &fakeGameAdmin{game: lobby.GameRecord{GameID: id}}
+	router, csrf := newGamesConsoleRouter(t, games, nil, nil)
+
+	rec := consolePost(t, router, "/_gm/games/"+id.String()+"/force-start", "_csrf="+csrf.Token("ops"))
+	if rec.Code != http.StatusSeeOther {
+		t.Fatalf("status = %d, want 303", rec.Code)
+	}
+	if games.forceStartCalls != 1 {
+		t.Errorf("AdminForceStart called %d times, want 1", games.forceStartCalls)
+	}
+}
+
+func TestConsoleGameForceStartRejectsBadCSRF(t *testing.T) {
+	id := uuid.New()
+	games := &fakeGameAdmin{game: lobby.GameRecord{GameID: id}}
+	router, _ := newGamesConsoleRouter(t, games, nil, nil)
+
+	rec := consolePost(t, router, "/_gm/games/"+id.String()+"/force-start", "")
+	if rec.Code != http.StatusForbidden {
+		t.Fatalf("status = %d, want 403", rec.Code)
+	}
+	if games.forceStartCalls != 0 {
+		t.Error("force-start must not run without a CSRF token")
+	}
+}
+
+func TestConsoleGameBanMember(t *testing.T) {
+	gameID := uuid.New()
+	target := uuid.New()
+	games := &fakeGameAdmin{game: lobby.GameRecord{GameID: gameID}}
+	router, csrf := newGamesConsoleRouter(t, games, nil, nil)
+
+	form := "_csrf=" + csrf.Token("ops") + "&user_id=" + target.String() + "&reason=cheating"
+	rec := consolePost(t, router, "/_gm/games/"+gameID.String()+"/ban-member", form)
+	if rec.Code != http.StatusSeeOther {
+		t.Fatalf("status = %d, want 303; body=%s", rec.Code, rec.Body.String())
+	}
+	if games.banCalls != 1 || games.lastBanUser != target || games.lastBanReason != "cheating" {
+		t.Errorf("ban recorded %d user=%s reason=%q", games.banCalls, games.lastBanUser, games.lastBanReason)
+	}
+}
+
+func TestConsoleGameBanMemberRejectsBadUUID(t *testing.T) {
+	gameID := uuid.New()
+	games := &fakeGameAdmin{game: lobby.GameRecord{GameID: gameID}}
+	router, csrf := newGamesConsoleRouter(t, games, nil, nil)
+
+	rec := consolePost(t, router, "/_gm/games/"+gameID.String()+"/ban-member", "_csrf="+csrf.Token("ops")+"&user_id=not-a-uuid")
+	if rec.Code != http.StatusBadRequest {
+		t.Fatalf("status = %d, want 400", rec.Code)
+	}
+	if games.banCalls != 0 {
+		t.Error("ban must not run with an invalid user id")
+	}
+}
+
+func TestConsoleRuntimePatch(t *testing.T) {
+	id := uuid.New()
+	games := &fakeGameAdmin{game: lobby.GameRecord{GameID: id}}
+	rt := &fakeRuntimeAdmin{}
+	router, csrf := newGamesConsoleRouter(t, games, rt, nil)
+
+	rec := consolePost(t, router, "/_gm/games/"+id.String()+"/runtime/patch", "_csrf="+csrf.Token("ops")+"&target_version=0.1.1")
+	if rec.Code != http.StatusSeeOther {
+		t.Fatalf("status = %d, want 303", rec.Code)
+	}
+	if rt.patchCalls != 1 || rt.lastPatchVersion != "0.1.1" {
+		t.Errorf("patch recorded %d version=%q", rt.patchCalls, rt.lastPatchVersion)
+	}
+}
+
+func TestConsoleRuntimePatchMissingVersion(t *testing.T) {
+	id := uuid.New()
+	games := &fakeGameAdmin{game: lobby.GameRecord{GameID: id}}
+	rt := &fakeRuntimeAdmin{}
+	router, csrf := newGamesConsoleRouter(t, games, rt, nil)
+
+	rec := consolePost(t, router, "/_gm/games/"+id.String()+"/runtime/patch", "_csrf="+csrf.Token("ops"))
+	if rec.Code != http.StatusBadRequest {
+		t.Fatalf("status = %d, want 400", rec.Code)
+	}
+	if rt.patchCalls != 0 {
+		t.Error("patch must not run without a target version")
+	}
+}
+
+func TestConsoleEngineVersions(t *testing.T) {
+	ev := &fakeEngineVersionAdmin{list: []runtime.EngineVersion{{Version: "0.1.0", ImageRef: "img:0.1.0", Enabled: true}}}
+	router, csrf := newGamesConsoleRouter(t, nil, nil, ev)
+
+	rec := consoleGet(t, router, "/_gm/engine-versions")
+	if rec.Code != http.StatusOK {
+		t.Fatalf("status = %d, want 200; body=%s", rec.Code, rec.Body.String())
+	}
+	for _, want := range []string{"0.1.0", "img:0.1.0", "Register version", "Disable"} {
+		if !strings.Contains(rec.Body.String(), want) {
+			t.Errorf("engine versions page missing %q", want)
+		}
+	}
+
+	rec = consolePost(t, router, "/_gm/engine-versions", "_csrf="+csrf.Token("ops")+"&version=0.2.0&image_ref=img:0.2.0&enabled=true")
+	if rec.Code != http.StatusSeeOther {
+		t.Fatalf("register status = %d, want 303", rec.Code)
+	}
+	if ev.registerCalls != 1 || ev.registered.Version != "0.2.0" || ev.registered.Enabled == nil || !*ev.registered.Enabled {
+		t.Errorf("register recorded %d version=%q enabled=%v", ev.registerCalls, ev.registered.Version, ev.registered.Enabled)
+	}
+
+	rec = consolePost(t, router, "/_gm/engine-versions/0.1.0/disable", "_csrf="+csrf.Token("ops"))
+	if rec.Code != http.StatusSeeOther {
+		t.Fatalf("disable status = %d, want 303", rec.Code)
+	}
+	if ev.disableCalls != 1 || ev.lastDisabled != "0.1.0" {
+		t.Errorf("disable recorded %d version=%q", ev.disableCalls, ev.lastDisabled)
+	}
+}
+
+func TestConsoleGamesUnavailable(t *testing.T) {
+	router, _ := newGamesConsoleRouter(t, nil, nil, nil)
+
+	rec := consoleGet(t, router, "/_gm/games")
+	if rec.Code != http.StatusServiceUnavailable {
+		t.Fatalf("status = %d, want 503", rec.Code)
+	}
+}

backend/internal/server/handlers_admin_console_mail.go

@@ -0,0 +1,327 @@
+package server
+
+import (
+	"context"
+	"errors"
+	"fmt"
+	"net/http"
+	"strings"
+
+	"galaxy/backend/internal/adminconsole"
+	"galaxy/backend/internal/diplomail"
+	"galaxy/backend/internal/mail"
+	"galaxy/backend/internal/notification"
+	"galaxy/backend/internal/server/clientip"
+	"galaxy/backend/internal/server/middleware/basicauth"
+
+	"github.com/gin-gonic/gin"
+	"github.com/google/uuid"
+	"go.uber.org/zap"
+)
+
+// MailAdmin is the subset of the mail service the console uses.
+type MailAdmin interface {
+	AdminListDeliveries(ctx context.Context, page, pageSize int) (mail.AdminListDeliveriesPage, error)
+	AdminGetDelivery(ctx context.Context, deliveryID uuid.UUID) (mail.Delivery, error)
+	AdminListAttempts(ctx context.Context, deliveryID uuid.UUID) ([]mail.Attempt, error)
+	AdminResendDelivery(ctx context.Context, deliveryID uuid.UUID) (mail.Delivery, error)
+	AdminListDeadLetters(ctx context.Context, page, pageSize int) (mail.AdminListDeadLettersPage, error)
+}
+
+// NotificationAdmin is the subset of the notification service the console uses.
+type NotificationAdmin interface {
+	AdminListNotifications(ctx context.Context, page, pageSize int) (notification.AdminListNotificationsPage, error)
+	AdminListDeadLetters(ctx context.Context, page, pageSize int) (notification.AdminListDeadLettersPage, error)
+	AdminListMalformed(ctx context.Context, page, pageSize int) (notification.AdminListMalformedPage, error)
+}
+
+// DiplomailAdmin is the subset of the diplomail service the console uses.
+type DiplomailAdmin interface {
+	SendAdminMultiGameBroadcast(ctx context.Context, in diplomail.SendMultiGameBroadcastInput) ([]diplomail.Message, int, error)
+}
+
+const consoleSnapshotPageSize = 50
+
+// MailPage renders GET /_gm/mail — paginated deliveries plus a dead-letter snapshot.
+func (h *AdminConsoleHandlers) MailPage() gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if h.mail == nil {
+			h.renderMessage(c, http.StatusServiceUnavailable, "mail", "Mail", "Mail administration is not available.", "bad", "/_gm/")
+			return
+		}
+		page := parsePositiveQueryInt(c.Query("page"), 1)
+		pageSize := parsePositiveQueryInt(c.Query("page_size"), 50)
+		ctx := c.Request.Context()
+
+		deliveries, err := h.mail.AdminListDeliveries(ctx, page, pageSize)
+		if err != nil {
+			h.logger.Error("admin console: list deliveries", zap.Error(err))
+			h.renderMessage(c, http.StatusInternalServerError, "mail", "Mail", "Failed to load deliveries.", "bad", "/_gm/")
+			return
+		}
+		dead, err := h.mail.AdminListDeadLetters(ctx, 1, consoleSnapshotPageSize)
+		if err != nil {
+			h.logger.Error("admin console: list mail dead-letters", zap.Error(err))
+			h.renderMessage(c, http.StatusInternalServerError, "mail", "Mail", "Failed to load dead-letters.", "bad", "/_gm/")
+			return
+		}
+		h.render(c, http.StatusOK, "mail", "mail", "Mail", toMailData(deliveries, dead))
+	}
+}
+
+// MailDeliveryDetail renders GET /_gm/mail/deliveries/:delivery_id.
+func (h *AdminConsoleHandlers) MailDeliveryDetail() gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if h.mail == nil {
+			h.renderMessage(c, http.StatusServiceUnavailable, "mail", "Mail", "Mail administration is not available.", "bad", "/_gm/")
+			return
+		}
+		deliveryID, ok := parseConsoleDeliveryID(c, h)
+		if !ok {
+			return
+		}
+		ctx := c.Request.Context()
+		delivery, err := h.mail.AdminGetDelivery(ctx, deliveryID)
+		if err != nil {
+			if errors.Is(err, mail.ErrDeliveryNotFound) {
+				h.renderMessage(c, http.StatusNotFound, "mail", "Delivery not found", "No such delivery.", "bad", "/_gm/mail")
+				return
+			}
+			h.logger.Error("admin console: get delivery", zap.Error(err))
+			h.renderMessage(c, http.StatusInternalServerError, "mail", "Mail", "Failed to load the delivery.", "bad", "/_gm/mail")
+			return
+		}
+		attempts, err := h.mail.AdminListAttempts(ctx, deliveryID)
+		if err != nil {
+			h.logger.Error("admin console: list attempts", zap.Error(err))
+			h.renderMessage(c, http.StatusInternalServerError, "mail", "Mail", "Failed to load attempts.", "bad", "/_gm/mail")
+			return
+		}
+		h.render(c, http.StatusOK, "mail_delivery", "mail", "Delivery", toMailDeliveryDetail(delivery, attempts))
+	}
+}
+
+// MailResend handles POST /_gm/mail/deliveries/:delivery_id/resend.
+func (h *AdminConsoleHandlers) MailResend() gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if h.mail == nil {
+			h.renderMessage(c, http.StatusServiceUnavailable, "mail", "Mail", "Mail administration is not available.", "bad", "/_gm/")
+			return
+		}
+		deliveryID, ok := parseConsoleDeliveryID(c, h)
+		if !ok {
+			return
+		}
+		back := "/_gm/mail/deliveries/" + deliveryID.String()
+		if _, err := h.mail.AdminResendDelivery(c.Request.Context(), deliveryID); err != nil {
+			h.logger.Error("admin console: resend delivery", zap.Error(err))
+			h.renderMessage(c, http.StatusInternalServerError, "mail", "Resend failed", "Failed to resend the delivery (it may already be sent).", "bad", back)
+			return
+		}
+		c.Redirect(http.StatusSeeOther, back)
+	}
+}
+
+// NotificationsPage renders GET /_gm/notifications — notifications, dead-letters,
+// and malformed intents on one overview page.
+func (h *AdminConsoleHandlers) NotificationsPage() gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if h.notifications == nil {
+			h.renderMessage(c, http.StatusServiceUnavailable, "mail", "Notifications", "Notification administration is not available.", "bad", "/_gm/")
+			return
+		}
+		ctx := c.Request.Context()
+		notifications, err := h.notifications.AdminListNotifications(ctx, 1, consoleSnapshotPageSize)
+		if err != nil {
+			h.logger.Error("admin console: list notifications", zap.Error(err))
+			h.renderMessage(c, http.StatusInternalServerError, "mail", "Notifications", "Failed to load notifications.", "bad", "/_gm/")
+			return
+		}
+		dead, err := h.notifications.AdminListDeadLetters(ctx, 1, consoleSnapshotPageSize)
+		if err != nil {
+			h.logger.Error("admin console: list notification dead-letters", zap.Error(err))
+			h.renderMessage(c, http.StatusInternalServerError, "mail", "Notifications", "Failed to load dead-letters.", "bad", "/_gm/")
+			return
+		}
+		malformed, err := h.notifications.AdminListMalformed(ctx, 1, consoleSnapshotPageSize)
+		if err != nil {
+			h.logger.Error("admin console: list malformed intents", zap.Error(err))
+			h.renderMessage(c, http.StatusInternalServerError, "mail", "Notifications", "Failed to load malformed intents.", "bad", "/_gm/")
+			return
+		}
+		h.render(c, http.StatusOK, "notifications", "mail", "Notifications", toNotificationsData(notifications, dead, malformed))
+	}
+}
+
+// BroadcastForm renders GET /_gm/broadcast.
+func (h *AdminConsoleHandlers) BroadcastForm() gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if h.diplomail == nil {
+			h.renderMessage(c, http.StatusServiceUnavailable, "mail", "Broadcast", "Broadcast is not available.", "bad", "/_gm/")
+			return
+		}
+		h.render(c, http.StatusOK, "broadcast", "mail", "Broadcast", nil)
+	}
+}
+
+// BroadcastSend handles POST /_gm/broadcast — multi-game admin broadcast.
+func (h *AdminConsoleHandlers) BroadcastSend() gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if h.diplomail == nil {
+			h.renderMessage(c, http.StatusServiceUnavailable, "mail", "Broadcast", "Broadcast is not available.", "bad", "/_gm/")
+			return
+		}
+		username, _ := basicauth.UsernameFromContext(c.Request.Context())
+		gameIDs, err := parseGameIDList(c.PostForm("game_ids"))
+		if err != nil {
+			h.renderMessage(c, http.StatusBadRequest, "mail", "Invalid input", "Game IDs must be valid UUIDs.", "bad", "/_gm/broadcast")
+			return
+		}
+		_, total, err := h.diplomail.SendAdminMultiGameBroadcast(c.Request.Context(), diplomail.SendMultiGameBroadcastInput{
+			CallerUsername: username,
+			Scope:          strings.TrimSpace(c.PostForm("scope")),
+			GameIDs:        gameIDs,
+			RecipientScope: strings.TrimSpace(c.PostForm("recipients")),
+			Subject:        strings.TrimSpace(c.PostForm("subject")),
+			Body:           c.PostForm("body"),
+			SenderIP:       clientip.ExtractSourceIP(c),
+		})
+		if err != nil {
+			if errors.Is(err, diplomail.ErrInvalidInput) {
+				h.renderMessage(c, http.StatusBadRequest, "mail", "Invalid input", "The broadcast was rejected: check the scope, recipients, and body.", "bad", "/_gm/broadcast")
+				return
+			}
+			h.logger.Error("admin console: broadcast", zap.Error(err))
+			h.renderMessage(c, http.StatusInternalServerError, "mail", "Broadcast failed", "Failed to send the broadcast.", "bad", "/_gm/broadcast")
+			return
+		}
+		h.renderMessage(c, http.StatusOK, "mail", "Broadcast sent", fmt.Sprintf("Broadcast delivered to %d recipients.", total), "ok", "/_gm/broadcast")
+	}
+}
+
+// parseConsoleDeliveryID parses the delivery_id path parameter, rendering a
+// console message page on failure.
+func parseConsoleDeliveryID(c *gin.Context, h *AdminConsoleHandlers) (uuid.UUID, bool) {
+	parsed, err := uuid.Parse(c.Param("delivery_id"))
+	if err != nil {
+		h.renderMessage(c, http.StatusBadRequest, "mail", "Invalid input", "delivery_id must be a valid UUID.", "bad", "/_gm/mail")
+		return uuid.Nil, false
+	}
+	return parsed, true
+}
+
+// parseGameIDList parses a comma-separated list of UUIDs, ignoring blanks.
+func parseGameIDList(raw string) ([]uuid.UUID, error) {
+	fields := strings.Split(raw, ",")
+	ids := make([]uuid.UUID, 0, len(fields))
+	for _, field := range fields {
+		field = strings.TrimSpace(field)
+		if field == "" {
+			continue
+		}
+		parsed, err := uuid.Parse(field)
+		if err != nil {
+			return nil, err
+		}
+		ids = append(ids, parsed)
+	}
+	return ids, nil
+}
+
+func toMailData(deliveries mail.AdminListDeliveriesPage, dead mail.AdminListDeadLettersPage) adminconsole.MailData {
+	data := adminconsole.MailData{
+		Deliveries:  make([]adminconsole.MailDeliveryRow, 0, len(deliveries.Items)),
+		DeadLetters: make([]adminconsole.MailDeadLetterRow, 0, len(dead.Items)),
+		Page:        deliveries.Page,
+		PageSize:    deliveries.PageSize,
+		Total:       deliveries.Total,
+		PrevPage:    deliveries.Page - 1,
+		NextPage:    deliveries.Page + 1,
+		HasPrev:     deliveries.Page > 1,
+		HasNext:     int64(deliveries.Page*deliveries.PageSize) < deliveries.Total,
+	}
+	for _, d := range deliveries.Items {
+		data.Deliveries = append(data.Deliveries, adminconsole.MailDeliveryRow{
+			DeliveryID:  d.DeliveryID.String(),
+			Template:    d.TemplateID,
+			Status:      d.Status,
+			Attempts:    d.Attempts,
+			NextAttempt: fmtConsoleTimePtr(d.NextAttemptAt),
+			Created:     fmtConsoleTime(d.CreatedAt),
+		})
+	}
+	for _, d := range dead.Items {
+		data.DeadLetters = append(data.DeadLetters, adminconsole.MailDeadLetterRow{
+			DeliveryID: d.DeliveryID.String(),
+			Reason:     d.Reason,
+			Archived:   fmtConsoleTime(d.ArchivedAt),
+		})
+	}
+	return data
+}
+
+func toMailDeliveryDetail(d mail.Delivery, attempts []mail.Attempt) adminconsole.MailDeliveryDetail {
+	detail := adminconsole.MailDeliveryDetail{
+		DeliveryID:   d.DeliveryID.String(),
+		Template:     d.TemplateID,
+		Status:       d.Status,
+		Attempts:     d.Attempts,
+		NextAttempt:  fmtConsoleTimePtr(d.NextAttemptAt),
+		LastError:    d.LastError,
+		Created:      fmtConsoleTime(d.CreatedAt),
+		Sent:         fmtConsoleTimePtr(d.SentAt),
+		DeadLettered: fmtConsoleTimePtr(d.DeadLetteredAt),
+		CanResend:    d.Status != mail.StatusSent,
+		AttemptRows:  make([]adminconsole.MailAttemptRow, 0, len(attempts)),
+	}
+	for _, a := range attempts {
+		detail.AttemptRows = append(detail.AttemptRows, adminconsole.MailAttemptRow{
+			AttemptNo: a.AttemptNo,
+			Outcome:   a.Outcome,
+			Started:   fmtConsoleTime(a.StartedAt),
+			Finished:  fmtConsoleTimePtr(a.FinishedAt),
+			Error:     a.Error,
+		})
+	}
+	return detail
+}
+
+func toNotificationsData(notifications notification.AdminListNotificationsPage, dead notification.AdminListDeadLettersPage, malformed notification.AdminListMalformedPage) adminconsole.NotificationsData {
+	data := adminconsole.NotificationsData{
+		Notifications: make([]adminconsole.NotificationRow, 0, len(notifications.Items)),
+		DeadLetters:   make([]adminconsole.NotificationDeadLetterRow, 0, len(dead.Items)),
+		Malformed:     make([]adminconsole.MalformedRow, 0, len(malformed.Items)),
+	}
+	for _, n := range notifications.Items {
+		data.Notifications = append(data.Notifications, adminconsole.NotificationRow{
+			NotificationID: n.NotificationID.String(),
+			Kind:           n.Kind,
+			UserID:         optionalUUID(n.UserID),
+			Created:        fmtConsoleTime(n.CreatedAt),
+		})
+	}
+	for _, d := range dead.Items {
+		data.DeadLetters = append(data.DeadLetters, adminconsole.NotificationDeadLetterRow{
+			NotificationID: d.NotificationID.String(),
+			RouteID:        d.RouteID.String(),
+			Reason:         d.Reason,
+			Archived:       fmtConsoleTime(d.ArchivedAt),
+		})
+	}
+	for _, m := range malformed.Items {
+		data.Malformed = append(data.Malformed, adminconsole.MalformedRow{
+			ID:       m.ID.String(),
+			Reason:   m.Reason,
+			Received: fmtConsoleTime(m.ReceivedAt),
+		})
+	}
+	return data
+}
+
+// optionalUUID renders a nullable user id; system-scoped rows have none.
+func optionalUUID(id *uuid.UUID) string {
+	if id == nil {
+		return "—"
+	}
+	return id.String()
+}

backend/internal/server/handlers_admin_console_mail_test.go

@@ -0,0 +1,242 @@
+package server
+
+import (
+	"context"
+	"net/http"
+	"strings"
+	"testing"
+	"time"
+
+	"galaxy/backend/internal/adminconsole"
+	"galaxy/backend/internal/diplomail"
+	"galaxy/backend/internal/mail"
+	"galaxy/backend/internal/notification"
+	"galaxy/backend/internal/server/middleware/basicauth"
+
+	"github.com/google/uuid"
+	"go.uber.org/zap"
+)
+
+type fakeMailAdmin struct {
+	deliveries  mail.AdminListDeliveriesPage
+	dead        mail.AdminListDeadLettersPage
+	delivery    mail.Delivery
+	getErr      error
+	attempts    []mail.Attempt
+	resendCalls int
+}
+
+func (f *fakeMailAdmin) AdminListDeliveries(context.Context, int, int) (mail.AdminListDeliveriesPage, error) {
+	return f.deliveries, nil
+}
+func (f *fakeMailAdmin) AdminGetDelivery(context.Context, uuid.UUID) (mail.Delivery, error) {
+	return f.delivery, f.getErr
+}
+func (f *fakeMailAdmin) AdminListAttempts(context.Context, uuid.UUID) ([]mail.Attempt, error) {
+	return f.attempts, nil
+}
+func (f *fakeMailAdmin) AdminResendDelivery(context.Context, uuid.UUID) (mail.Delivery, error) {
+	f.resendCalls++
+	return f.delivery, nil
+}
+func (f *fakeMailAdmin) AdminListDeadLetters(context.Context, int, int) (mail.AdminListDeadLettersPage, error) {
+	return f.dead, nil
+}
+
+type fakeNotificationAdmin struct {
+	notifications notification.AdminListNotificationsPage
+	dead          notification.AdminListDeadLettersPage
+	malformed     notification.AdminListMalformedPage
+}
+
+func (f *fakeNotificationAdmin) AdminListNotifications(context.Context, int, int) (notification.AdminListNotificationsPage, error) {
+	return f.notifications, nil
+}
+func (f *fakeNotificationAdmin) AdminListDeadLetters(context.Context, int, int) (notification.AdminListDeadLettersPage, error) {
+	return f.dead, nil
+}
+func (f *fakeNotificationAdmin) AdminListMalformed(context.Context, int, int) (notification.AdminListMalformedPage, error) {
+	return f.malformed, nil
+}
+
+type fakeDiplomailAdmin struct {
+	total          int
+	err            error
+	broadcastCalls int
+	last           diplomail.SendMultiGameBroadcastInput
+}
+
+func (f *fakeDiplomailAdmin) SendAdminMultiGameBroadcast(_ context.Context, in diplomail.SendMultiGameBroadcastInput) ([]diplomail.Message, int, error) {
+	f.broadcastCalls++
+	f.last = in
+	if f.err != nil {
+		return nil, 0, f.err
+	}
+	return nil, f.total, nil
+}
+
+func mailConsoleRouter(t *testing.T, m MailAdmin, n NotificationAdmin, d DiplomailAdmin) (http.Handler, *adminconsole.CSRF) {
+	t.Helper()
+	csrf := adminconsole.NewCSRF([]byte("test-key"))
+	handler, err := NewRouter(RouterDependencies{
+		Logger:        zap.NewNop(),
+		AdminVerifier: basicauth.NewStaticVerifier("secret"),
+		AdminConsole:  NewAdminConsoleHandlers(AdminConsoleDeps{CSRF: csrf, Mail: m, Notifications: n, Diplomail: d}),
+	})
+	if err != nil {
+		t.Fatalf("NewRouter: %v", err)
+	}
+	return handler, csrf
+}
+
+func TestConsoleMailPage(t *testing.T) {
+	id := uuid.New()
+	m := &fakeMailAdmin{
+		deliveries: mail.AdminListDeliveriesPage{
+			Items: []mail.Delivery{{DeliveryID: id, TemplateID: "auth.login_code", Status: "pending", CreatedAt: time.Now()}},
+			Page:  1, PageSize: 50, Total: 1,
+		},
+		dead: mail.AdminListDeadLettersPage{
+			Items: []mail.DeadLetter{{DeliveryID: uuid.New(), Reason: "smtp 550", ArchivedAt: time.Now()}},
+		},
+	}
+	router, _ := mailConsoleRouter(t, m, nil, nil)
+
+	rec := consoleGet(t, router, "/_gm/mail")
+	if rec.Code != http.StatusOK {
+		t.Fatalf("status = %d, want 200; body=%s", rec.Code, rec.Body.String())
+	}
+	for _, want := range []string{"auth.login_code", "pending", "Dead-letters", "smtp 550"} {
+		if !strings.Contains(rec.Body.String(), want) {
+			t.Errorf("mail page missing %q", want)
+		}
+	}
+}
+
+func TestConsoleMailDeliveryDetail(t *testing.T) {
+	id := uuid.New()
+	m := &fakeMailAdmin{
+		delivery: mail.Delivery{DeliveryID: id, TemplateID: "auth.login_code", Status: "pending", Attempts: 2},
+		attempts: []mail.Attempt{{AttemptNo: 1, Outcome: "transient_failure", StartedAt: time.Now(), Error: "timeout"}},
+	}
+	router, _ := mailConsoleRouter(t, m, nil, nil)
+
+	rec := consoleGet(t, router, "/_gm/mail/deliveries/"+id.String())
+	if rec.Code != http.StatusOK {
+		t.Fatalf("status = %d, want 200; body=%s", rec.Code, rec.Body.String())
+	}
+	for _, want := range []string{id.String(), "auth.login_code", "Attempts", "transient_failure", "Resend"} {
+		if !strings.Contains(rec.Body.String(), want) {
+			t.Errorf("delivery detail missing %q", want)
+		}
+	}
+}
+
+func TestConsoleMailDeliveryDetailNotFound(t *testing.T) {
+	m := &fakeMailAdmin{getErr: mail.ErrDeliveryNotFound}
+	router, _ := mailConsoleRouter(t, m, nil, nil)
+
+	rec := consoleGet(t, router, "/_gm/mail/deliveries/"+uuid.New().String())
+	if rec.Code != http.StatusNotFound {
+		t.Fatalf("status = %d, want 404", rec.Code)
+	}
+}
+
+func TestConsoleMailResend(t *testing.T) {
+	id := uuid.New()
+	m := &fakeMailAdmin{delivery: mail.Delivery{DeliveryID: id}}
+	router, csrf := mailConsoleRouter(t, m, nil, nil)
+
+	rec := consolePost(t, router, "/_gm/mail/deliveries/"+id.String()+"/resend", "_csrf="+csrf.Token("ops"))
+	if rec.Code != http.StatusSeeOther {
+		t.Fatalf("status = %d, want 303", rec.Code)
+	}
+	if m.resendCalls != 1 {
+		t.Errorf("AdminResendDelivery called %d times, want 1", m.resendCalls)
+	}
+}
+
+func TestConsoleMailResendRejectsBadCSRF(t *testing.T) {
+	id := uuid.New()
+	m := &fakeMailAdmin{delivery: mail.Delivery{DeliveryID: id}}
+	router, _ := mailConsoleRouter(t, m, nil, nil)
+
+	rec := consolePost(t, router, "/_gm/mail/deliveries/"+id.String()+"/resend", "")
+	if rec.Code != http.StatusForbidden {
+		t.Fatalf("status = %d, want 403", rec.Code)
+	}
+	if m.resendCalls != 0 {
+		t.Error("resend must not run without a CSRF token")
+	}
+}
+
+func TestConsoleNotificationsPage(t *testing.T) {
+	n := &fakeNotificationAdmin{
+		notifications: notification.AdminListNotificationsPage{Items: []notification.Notification{{NotificationID: uuid.New(), Kind: "lobby.invite.received"}}},
+		dead:          notification.AdminListDeadLettersPage{Items: []notification.DeadLetter{{NotificationID: uuid.New(), RouteID: uuid.New(), Reason: "push gone"}}},
+		malformed:     notification.AdminListMalformedPage{Items: []notification.MalformedIntent{{ID: uuid.New(), Reason: "bad shape"}}},
+	}
+	router, _ := mailConsoleRouter(t, nil, n, nil)
+
+	rec := consoleGet(t, router, "/_gm/notifications")
+	if rec.Code != http.StatusOK {
+		t.Fatalf("status = %d, want 200; body=%s", rec.Code, rec.Body.String())
+	}
+	for _, want := range []string{"lobby.invite.received", "push gone", "bad shape", "Malformed intents"} {
+		if !strings.Contains(rec.Body.String(), want) {
+			t.Errorf("notifications page missing %q", want)
+		}
+	}
+}
+
+func TestConsoleBroadcastForm(t *testing.T) {
+	router, _ := mailConsoleRouter(t, nil, nil, &fakeDiplomailAdmin{})
+
+	rec := consoleGet(t, router, "/_gm/broadcast")
+	if rec.Code != http.StatusOK {
+		t.Fatalf("status = %d, want 200", rec.Code)
+	}
+	if !strings.Contains(rec.Body.String(), "Send broadcast") {
+		t.Error("broadcast form missing")
+	}
+}
+
+func TestConsoleBroadcastSend(t *testing.T) {
+	d := &fakeDiplomailAdmin{total: 5}
+	router, csrf := mailConsoleRouter(t, nil, nil, d)
+
+	form := "_csrf=" + csrf.Token("ops") + "&scope=all_running&recipients=active&body=hello"
+	rec := consolePost(t, router, "/_gm/broadcast", form)
+	if rec.Code != http.StatusOK {
+		t.Fatalf("status = %d, want 200; body=%s", rec.Code, rec.Body.String())
+	}
+	if !strings.Contains(rec.Body.String(), "5 recipients") {
+		t.Errorf("broadcast result missing recipient count; body=%s", rec.Body.String())
+	}
+	if d.broadcastCalls != 1 || d.last.Scope != "all_running" || d.last.Body != "hello" || d.last.CallerUsername != "ops" {
+		t.Errorf("broadcast input = %+v (calls=%d)", d.last, d.broadcastCalls)
+	}
+}
+
+func TestConsoleBroadcastSendBadGameIDs(t *testing.T) {
+	d := &fakeDiplomailAdmin{}
+	router, csrf := mailConsoleRouter(t, nil, nil, d)
+
+	form := "_csrf=" + csrf.Token("ops") + "&scope=selected&game_ids=not-a-uuid&body=hello"
+	rec := consolePost(t, router, "/_gm/broadcast", form)
+	if rec.Code != http.StatusBadRequest {
+		t.Fatalf("status = %d, want 400", rec.Code)
+	}
+	if d.broadcastCalls != 0 {
+		t.Error("broadcast must not run with invalid game ids")
+	}
+}
+
+func TestConsoleMailUnavailable(t *testing.T) {
+	router, _ := mailConsoleRouter(t, nil, nil, nil)
+
+	rec := consoleGet(t, router, "/_gm/mail")
+	if rec.Code != http.StatusServiceUnavailable {
+		t.Fatalf("status = %d, want 503", rec.Code)
+	}
+}

backend/internal/server/handlers_admin_console_operators.go

@@ -0,0 +1,149 @@
+package server
+
+import (
+	"context"
+	"errors"
+	"net/http"
+	"strings"
+
+	"galaxy/backend/internal/admin"
+	"galaxy/backend/internal/adminconsole"
+
+	"github.com/gin-gonic/gin"
+	"go.uber.org/zap"
+)
+
+// OperatorAdmin is the subset of the admin-account service the console uses.
+// *admin.Service satisfies it.
+type OperatorAdmin interface {
+	List(ctx context.Context) ([]admin.Admin, error)
+	Create(ctx context.Context, in admin.CreateInput) (admin.Admin, error)
+	Disable(ctx context.Context, username string) (admin.Admin, error)
+	Enable(ctx context.Context, username string) (admin.Admin, error)
+	ResetPassword(ctx context.Context, username, password string) (admin.Admin, error)
+}
+
+// OperatorsList renders GET /_gm/operators.
+func (h *AdminConsoleHandlers) OperatorsList() gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if h.operators == nil {
+			h.renderMessage(c, http.StatusServiceUnavailable, "operators", "Operators", "Operator administration is not available.", "bad", "/_gm/")
+			return
+		}
+		admins, err := h.operators.List(c.Request.Context())
+		if err != nil {
+			h.logger.Error("admin console: list operators", zap.Error(err))
+			h.renderMessage(c, http.StatusInternalServerError, "operators", "Operators", "Failed to load operators.", "bad", "/_gm/")
+			return
+		}
+		h.render(c, http.StatusOK, "operators", "operators", "Operators", toOperatorsData(admins))
+	}
+}
+
+// OperatorCreate handles POST /_gm/operators.
+func (h *AdminConsoleHandlers) OperatorCreate() gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if h.operators == nil {
+			h.renderMessage(c, http.StatusServiceUnavailable, "operators", "Operators", "Operator administration is not available.", "bad", "/_gm/")
+			return
+		}
+		_, err := h.operators.Create(c.Request.Context(), admin.CreateInput{
+			Username: strings.TrimSpace(c.PostForm("username")),
+			Password: c.PostForm("password"),
+		})
+		if err != nil {
+			switch {
+			case errors.Is(err, admin.ErrUsernameTaken):
+				h.renderMessage(c, http.StatusConflict, "operators", "Username taken", "That username is already in use.", "bad", "/_gm/operators")
+			case errors.Is(err, admin.ErrInvalidInput):
+				h.renderMessage(c, http.StatusBadRequest, "operators", "Invalid input", "Username and password are required.", "bad", "/_gm/operators")
+			default:
+				h.logger.Error("admin console: create operator", zap.Error(err))
+				h.renderMessage(c, http.StatusInternalServerError, "operators", "Create failed", "Failed to create the operator.", "bad", "/_gm/operators")
+			}
+			return
+		}
+		c.Redirect(http.StatusSeeOther, "/_gm/operators")
+	}
+}
+
+// OperatorDisable handles POST /_gm/operators/:username/disable.
+func (h *AdminConsoleHandlers) OperatorDisable() gin.HandlerFunc {
+	return h.operatorAction("disable", func(ctx context.Context, username string) error {
+		_, err := h.operators.Disable(ctx, username)
+		return err
+	})
+}
+
+// OperatorEnable handles POST /_gm/operators/:username/enable.
+func (h *AdminConsoleHandlers) OperatorEnable() gin.HandlerFunc {
+	return h.operatorAction("enable", func(ctx context.Context, username string) error {
+		_, err := h.operators.Enable(ctx, username)
+		return err
+	})
+}
+
+// OperatorResetPassword handles POST /_gm/operators/:username/reset-password.
+func (h *AdminConsoleHandlers) OperatorResetPassword() gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if h.operators == nil {
+			h.renderMessage(c, http.StatusServiceUnavailable, "operators", "Operators", "Operator administration is not available.", "bad", "/_gm/")
+			return
+		}
+		username := c.Param("username")
+		password := c.PostForm("password")
+		if strings.TrimSpace(password) == "" {
+			h.renderMessage(c, http.StatusBadRequest, "operators", "Invalid input", "A new password is required.", "bad", "/_gm/operators")
+			return
+		}
+		if _, err := h.operators.ResetPassword(c.Request.Context(), username, password); err != nil {
+			if errors.Is(err, admin.ErrNotFound) {
+				h.renderMessage(c, http.StatusNotFound, "operators", "Operator not found", "No such operator.", "bad", "/_gm/operators")
+				return
+			}
+			if errors.Is(err, admin.ErrInvalidInput) {
+				h.renderMessage(c, http.StatusBadRequest, "operators", "Invalid input", "The password was rejected.", "bad", "/_gm/operators")
+				return
+			}
+			h.logger.Error("admin console: reset operator password", zap.Error(err))
+			h.renderMessage(c, http.StatusInternalServerError, "operators", "Reset failed", "Failed to reset the password.", "bad", "/_gm/operators")
+			return
+		}
+		c.Redirect(http.StatusSeeOther, "/_gm/operators")
+	}
+}
+
+// operatorAction is the shared shape for operator POST actions that take only
+// the username and redirect back to the list.
+func (h *AdminConsoleHandlers) operatorAction(label string, run func(context.Context, string) error) gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if h.operators == nil {
+			h.renderMessage(c, http.StatusServiceUnavailable, "operators", "Operators", "Operator administration is not available.", "bad", "/_gm/")
+			return
+		}
+		if err := run(c.Request.Context(), c.Param("username")); err != nil {
+			if errors.Is(err, admin.ErrNotFound) {
+				h.renderMessage(c, http.StatusNotFound, "operators", "Operator not found", "No such operator.", "bad", "/_gm/operators")
+				return
+			}
+			h.logger.Error("admin console: operator "+label, zap.Error(err))
+			h.renderMessage(c, http.StatusInternalServerError, "operators", "Action failed", "The "+label+" action failed.", "bad", "/_gm/operators")
+			return
+		}
+		c.Redirect(http.StatusSeeOther, "/_gm/operators")
+	}
+}
+
+// toOperatorsData maps admin accounts into the operators view model.
+func toOperatorsData(admins []admin.Admin) adminconsole.OperatorsData {
+	data := adminconsole.OperatorsData{Items: make([]adminconsole.OperatorRow, 0, len(admins))}
+	for _, a := range admins {
+		data.Items = append(data.Items, adminconsole.OperatorRow{
+			Username:   a.Username,
+			CreatedAt:  fmtConsoleTime(a.CreatedAt),
+			LastUsedAt: fmtConsoleTimePtr(a.LastUsedAt),
+			Disabled:   a.DisabledAt != nil,
+		})
+	}
+	return data
+}

backend/internal/server/handlers_admin_console_operators_test.go

@@ -0,0 +1,166 @@
+package server
+
+import (
+	"context"
+	"net/http"
+	"strings"
+	"testing"
+
+	"galaxy/backend/internal/admin"
+	"galaxy/backend/internal/adminconsole"
+	"galaxy/backend/internal/server/middleware/basicauth"
+
+	"go.uber.org/zap"
+)
+
+type fakeOperatorAdmin struct {
+	list      []admin.Admin
+	createErr error
+
+	created       admin.CreateInput
+	createCalls   int
+	disableCalls  int
+	enableCalls   int
+	resetCalls    int
+	lastResetUser string
+	lastResetPass string
+}
+
+func (f *fakeOperatorAdmin) List(context.Context) ([]admin.Admin, error) { return f.list, nil }
+func (f *fakeOperatorAdmin) Create(_ context.Context, in admin.CreateInput) (admin.Admin, error) {
+	f.createCalls++
+	f.created = in
+	if f.createErr != nil {
+		return admin.Admin{}, f.createErr
+	}
+	return admin.Admin{Username: in.Username}, nil
+}
+func (f *fakeOperatorAdmin) Disable(_ context.Context, username string) (admin.Admin, error) {
+	f.disableCalls++
+	return admin.Admin{Username: username}, nil
+}
+func (f *fakeOperatorAdmin) Enable(_ context.Context, username string) (admin.Admin, error) {
+	f.enableCalls++
+	return admin.Admin{Username: username}, nil
+}
+func (f *fakeOperatorAdmin) ResetPassword(_ context.Context, username, password string) (admin.Admin, error) {
+	f.resetCalls++
+	f.lastResetUser = username
+	f.lastResetPass = password
+	return admin.Admin{Username: username}, nil
+}
+
+func operatorsRouter(t *testing.T, operators OperatorAdmin) (http.Handler, *adminconsole.CSRF) {
+	t.Helper()
+	csrf := adminconsole.NewCSRF([]byte("test-key"))
+	handler, err := NewRouter(RouterDependencies{
+		Logger:        zap.NewNop(),
+		AdminVerifier: basicauth.NewStaticVerifier("secret"),
+		AdminConsole:  NewAdminConsoleHandlers(AdminConsoleDeps{CSRF: csrf, Operators: operators}),
+	})
+	if err != nil {
+		t.Fatalf("NewRouter: %v", err)
+	}
+	return handler, csrf
+}
+
+func TestConsoleOperatorsList(t *testing.T) {
+	fake := &fakeOperatorAdmin{list: []admin.Admin{{Username: "root"}}}
+	router, _ := operatorsRouter(t, fake)
+
+	rec := consoleGet(t, router, "/_gm/operators")
+	if rec.Code != http.StatusOK {
+		t.Fatalf("status = %d, want 200; body=%s", rec.Code, rec.Body.String())
+	}
+	for _, want := range []string{"root", "Create operator", "Reset"} {
+		if !strings.Contains(rec.Body.String(), want) {
+			t.Errorf("operators page missing %q", want)
+		}
+	}
+}
+
+func TestConsoleOperatorCreate(t *testing.T) {
+	fake := &fakeOperatorAdmin{}
+	router, csrf := operatorsRouter(t, fake)
+
+	rec := consolePost(t, router, "/_gm/operators", "_csrf="+csrf.Token("ops")+"&username=mod&password=s3cret")
+	if rec.Code != http.StatusSeeOther {
+		t.Fatalf("status = %d, want 303; body=%s", rec.Code, rec.Body.String())
+	}
+	if fake.createCalls != 1 || fake.created.Username != "mod" || fake.created.Password != "s3cret" {
+		t.Errorf("create recorded %d username=%q", fake.createCalls, fake.created.Username)
+	}
+}
+
+func TestConsoleOperatorCreateConflict(t *testing.T) {
+	fake := &fakeOperatorAdmin{createErr: admin.ErrUsernameTaken}
+	router, csrf := operatorsRouter(t, fake)
+
+	rec := consolePost(t, router, "/_gm/operators", "_csrf="+csrf.Token("ops")+"&username=root&password=x")
+	if rec.Code != http.StatusConflict {
+		t.Fatalf("status = %d, want 409", rec.Code)
+	}
+}
+
+func TestConsoleOperatorDisableEnable(t *testing.T) {
+	fake := &fakeOperatorAdmin{}
+	router, csrf := operatorsRouter(t, fake)
+
+	if rec := consolePost(t, router, "/_gm/operators/root/disable", "_csrf="+csrf.Token("ops")); rec.Code != http.StatusSeeOther {
+		t.Fatalf("disable status = %d, want 303", rec.Code)
+	}
+	if rec := consolePost(t, router, "/_gm/operators/root/enable", "_csrf="+csrf.Token("ops")); rec.Code != http.StatusSeeOther {
+		t.Fatalf("enable status = %d, want 303", rec.Code)
+	}
+	if fake.disableCalls != 1 || fake.enableCalls != 1 {
+		t.Errorf("disable=%d enable=%d, want 1/1", fake.disableCalls, fake.enableCalls)
+	}
+}
+
+func TestConsoleOperatorResetPassword(t *testing.T) {
+	fake := &fakeOperatorAdmin{}
+	router, csrf := operatorsRouter(t, fake)
+
+	rec := consolePost(t, router, "/_gm/operators/root/reset-password", "_csrf="+csrf.Token("ops")+"&password=newpass")
+	if rec.Code != http.StatusSeeOther {
+		t.Fatalf("status = %d, want 303", rec.Code)
+	}
+	if fake.resetCalls != 1 || fake.lastResetUser != "root" || fake.lastResetPass != "newpass" {
+		t.Errorf("reset recorded %d user=%q", fake.resetCalls, fake.lastResetUser)
+	}
+}
+
+func TestConsoleOperatorResetPasswordMissing(t *testing.T) {
+	fake := &fakeOperatorAdmin{}
+	router, csrf := operatorsRouter(t, fake)
+
+	rec := consolePost(t, router, "/_gm/operators/root/reset-password", "_csrf="+csrf.Token("ops"))
+	if rec.Code != http.StatusBadRequest {
+		t.Fatalf("status = %d, want 400", rec.Code)
+	}
+	if fake.resetCalls != 0 {
+		t.Error("reset must not run without a password")
+	}
+}
+
+func TestConsoleOperatorRejectsBadCSRF(t *testing.T) {
+	fake := &fakeOperatorAdmin{}
+	router, _ := operatorsRouter(t, fake)
+
+	rec := consolePost(t, router, "/_gm/operators/root/disable", "")
+	if rec.Code != http.StatusForbidden {
+		t.Fatalf("status = %d, want 403", rec.Code)
+	}
+	if fake.disableCalls != 0 {
+		t.Error("disable must not run without a CSRF token")
+	}
+}
+
+func TestConsoleOperatorsUnavailable(t *testing.T) {
+	router, _ := operatorsRouter(t, nil)
+
+	rec := consoleGet(t, router, "/_gm/operators")
+	if rec.Code != http.StatusServiceUnavailable {
+		t.Fatalf("status = %d, want 503", rec.Code)
+	}
+}

backend/internal/server/handlers_admin_console_test.go

@@ -0,0 +1,214 @@
+package server
+
+import (
+	"context"
+	"io"
+	"net/http"
+	"net/http/httptest"
+	"strings"
+	"testing"
+
+	"galaxy/backend/internal/adminconsole"
+	"galaxy/backend/internal/opsstatus"
+	"galaxy/backend/internal/server/middleware/basicauth"
+
+	"github.com/gin-gonic/gin"
+	"go.uber.org/zap"
+)
+
+// fakeMonitor is a static opsstatus.Reader for dashboard rendering tests.
+type fakeMonitor struct {
+	snapshot opsstatus.Snapshot
+}
+
+func (f fakeMonitor) Collect(context.Context) opsstatus.Snapshot {
+	return f.snapshot
+}
+
+func newConsoleTestRouter(t *testing.T) http.Handler {
+	t.Helper()
+	handler, err := NewRouter(RouterDependencies{
+		Logger:        zap.NewNop(),
+		AdminVerifier: basicauth.NewStaticVerifier("secret"),
+		AdminConsole:  NewAdminConsoleHandlers(AdminConsoleDeps{CSRF: adminconsole.NewCSRF([]byte("test-key"))}),
+	})
+	if err != nil {
+		t.Fatalf("NewRouter: %v", err)
+	}
+	return handler
+}
+
+func TestAdminConsoleRequiresAuth(t *testing.T) {
+	router := newConsoleTestRouter(t)
+
+	req := httptest.NewRequest(http.MethodGet, "/_gm/", nil)
+	rec := httptest.NewRecorder()
+	router.ServeHTTP(rec, req)
+
+	if rec.Code != http.StatusUnauthorized {
+		t.Fatalf("status = %d, want 401", rec.Code)
+	}
+	if got := rec.Header().Get("WWW-Authenticate"); !strings.Contains(got, "Basic") {
+		t.Fatalf("WWW-Authenticate = %q, want a Basic challenge", got)
+	}
+}
+
+func TestAdminConsoleDashboardRenders(t *testing.T) {
+	router := newConsoleTestRouter(t)
+
+	for _, path := range []string{"/_gm", "/_gm/"} {
+		req := httptest.NewRequest(http.MethodGet, path, nil)
+		req.SetBasicAuth("ops", "secret")
+		rec := httptest.NewRecorder()
+		router.ServeHTTP(rec, req)
+
+		if rec.Code != http.StatusOK {
+			t.Fatalf("GET %s status = %d, want 200; body=%s", path, rec.Code, rec.Body.String())
+		}
+		if ct := rec.Header().Get("Content-Type"); !strings.HasPrefix(ct, "text/html") {
+			t.Errorf("GET %s content-type = %q, want text/html", path, ct)
+		}
+		body := rec.Body.String()
+		if !strings.Contains(body, "Dashboard") {
+			t.Errorf("GET %s body missing the dashboard heading", path)
+		}
+		if !strings.Contains(body, "ops") {
+			t.Errorf("GET %s body missing the operator name", path)
+		}
+	}
+}
+
+func TestAdminConsoleDashboardShowsMonitoring(t *testing.T) {
+	monitor := fakeMonitor{snapshot: opsstatus.Snapshot{
+		PostgresHealthy:       true,
+		Runtimes:              []opsstatus.StatusCount{{Status: "running", Count: 3}, {Status: "stopped", Count: 1}},
+		MailDeliveries:        []opsstatus.StatusCount{{Status: "pending", Count: 2}},
+		NotificationRoutes:    []opsstatus.StatusCount{{Status: "published", Count: 9}},
+		NotificationMalformed: 4,
+		Errors:                []string{"notification route counts: boom"},
+	}}
+
+	handler, err := NewRouter(RouterDependencies{
+		Logger:        zap.NewNop(),
+		AdminVerifier: basicauth.NewStaticVerifier("secret"),
+		AdminConsole: NewAdminConsoleHandlers(AdminConsoleDeps{
+			CSRF:    adminconsole.NewCSRF([]byte("test-key")),
+			Monitor: monitor,
+			Ready:   func() bool { return true },
+		}),
+	})
+	if err != nil {
+		t.Fatalf("NewRouter: %v", err)
+	}
+
+	req := httptest.NewRequest(http.MethodGet, "/_gm/", nil)
+	req.SetBasicAuth("ops", "secret")
+	rec := httptest.NewRecorder()
+	handler.ServeHTTP(rec, req)
+
+	if rec.Code != http.StatusOK {
+		t.Fatalf("status = %d, want 200; body=%s", rec.Code, rec.Body.String())
+	}
+	body := rec.Body.String()
+	for _, want := range []string{
+		"Game runtimes", "running", "stopped",
+		"Mail deliveries", "pending",
+		"Notification routes", "published",
+		"Malformed notifications",
+		"notification route counts: boom",
+		"healthy",
+	} {
+		if !strings.Contains(body, want) {
+			t.Errorf("dashboard body missing %q", want)
+		}
+	}
+}
+
+func TestAdminConsoleDashboardWithoutMonitor(t *testing.T) {
+	router := newConsoleTestRouter(t) // no monitor wired
+
+	req := httptest.NewRequest(http.MethodGet, "/_gm/", nil)
+	req.SetBasicAuth("ops", "secret")
+	rec := httptest.NewRecorder()
+	router.ServeHTTP(rec, req)
+
+	if rec.Code != http.StatusOK {
+		t.Fatalf("status = %d, want 200", rec.Code)
+	}
+	if !strings.Contains(rec.Body.String(), "Monitoring is not wired") {
+		t.Error("dashboard without a monitor should note that monitoring is unavailable")
+	}
+}
+
+func TestAdminConsoleServesAsset(t *testing.T) {
+	router := newConsoleTestRouter(t)
+
+	req := httptest.NewRequest(http.MethodGet, "/_gm/assets/console.css", nil)
+	req.SetBasicAuth("ops", "secret")
+	rec := httptest.NewRecorder()
+	router.ServeHTTP(rec, req)
+
+	if rec.Code != http.StatusOK {
+		t.Fatalf("asset status = %d, want 200", rec.Code)
+	}
+	if ct := rec.Header().Get("Content-Type"); !strings.Contains(ct, "text/css") {
+		t.Errorf("asset content-type = %q, want text/css", ct)
+	}
+}
+
+func TestAdminConsoleRequireCSRF(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+
+	csrf := adminconsole.NewCSRF([]byte("test-key"))
+	console := NewAdminConsoleHandlers(AdminConsoleDeps{CSRF: csrf})
+
+	engine := gin.New()
+	engine.Use(func(c *gin.Context) {
+		c.Request = c.Request.WithContext(basicauth.WithUsername(c.Request.Context(), "ops"))
+		c.Next()
+	})
+	engine.Use(console.RequireCSRF())
+	engine.GET("/x", func(c *gin.Context) { c.Status(http.StatusOK) })
+	engine.POST("/x", func(c *gin.Context) { c.Status(http.StatusOK) })
+
+	token := csrf.Token("ops")
+
+	cases := []struct {
+		name   string
+		method string
+		form   string
+		origin string
+		host   string
+		want   int
+	}{
+		{"get is a safe method", http.MethodGet, "", "", "galaxy.lan", http.StatusOK},
+		{"valid token, same origin", http.MethodPost, "_csrf=" + token, "https://galaxy.lan", "galaxy.lan", http.StatusOK},
+		{"valid token, no origin header", http.MethodPost, "_csrf=" + token, "", "galaxy.lan", http.StatusOK},
+		{"missing token", http.MethodPost, "", "https://galaxy.lan", "galaxy.lan", http.StatusForbidden},
+		{"wrong token", http.MethodPost, "_csrf=bogus", "https://galaxy.lan", "galaxy.lan", http.StatusForbidden},
+		{"cross-origin", http.MethodPost, "_csrf=" + token, "https://evil.example", "galaxy.lan", http.StatusForbidden},
+	}
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			var body io.Reader
+			if tc.form != "" {
+				body = strings.NewReader(tc.form)
+			}
+			req := httptest.NewRequest(tc.method, "/x", body)
+			if tc.form != "" {
+				req.Header.Set("Content-Type", "application/x-www-form-urlencoded")
+			}
+			if tc.origin != "" {
+				req.Header.Set("Origin", tc.origin)
+			}
+			req.Host = tc.host
+
+			rec := httptest.NewRecorder()
+			engine.ServeHTTP(rec, req)
+
+			if rec.Code != tc.want {
+				t.Fatalf("status = %d, want %d (body=%s)", rec.Code, tc.want, rec.Body.String())
+			}
+		})
+	}
+}

backend/internal/server/handlers_admin_console_users.go

@@ -0,0 +1,252 @@
+package server
+
+import (
+	"context"
+	"errors"
+	"net/http"
+	"strings"
+	"time"
+
+	"galaxy/backend/internal/adminconsole"
+	"galaxy/backend/internal/server/middleware/basicauth"
+	"galaxy/backend/internal/user"
+
+	"github.com/gin-gonic/gin"
+	"github.com/google/uuid"
+	"go.uber.org/zap"
+)
+
+// UserAdmin is the subset of the user service the operator console depends on.
+// *user.Service satisfies it; tests supply a fake so the console pages render
+// without a database.
+type UserAdmin interface {
+	ListAccounts(ctx context.Context, page, pageSize int) (user.AccountPage, error)
+	GetAccount(ctx context.Context, userID uuid.UUID) (user.Account, error)
+	ApplySanction(ctx context.Context, input user.ApplySanctionInput) (user.Account, error)
+	ApplyEntitlement(ctx context.Context, input user.ApplyEntitlementInput) (user.Account, error)
+	SoftDelete(ctx context.Context, userID uuid.UUID, actor user.ActorRef) error
+}
+
+// consoleTiers lists the selectable entitlement tiers in display order.
+var consoleTiers = []string{user.TierFree, user.TierMonthly, user.TierYearly, user.TierPermanent}
+
+// UsersList renders GET /_gm/users — the paginated account list.
+func (h *AdminConsoleHandlers) UsersList() gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if h.users == nil {
+			h.renderMessage(c, http.StatusServiceUnavailable, "users", "Users", "User administration is not available.", "bad", "/_gm/")
+			return
+		}
+		page := parsePositiveQueryInt(c.Query("page"), 1)
+		pageSize := parsePositiveQueryInt(c.Query("page_size"), 50)
+
+		result, err := h.users.ListAccounts(c.Request.Context(), page, pageSize)
+		if err != nil {
+			h.logger.Error("admin console: list users", zap.Error(err))
+			h.renderMessage(c, http.StatusInternalServerError, "users", "Users", "Failed to load users.", "bad", "/_gm/")
+			return
+		}
+		h.render(c, http.StatusOK, "users", "users", "Users", toUsersListData(result))
+	}
+}
+
+// UserDetail renders GET /_gm/users/:user_id.
+func (h *AdminConsoleHandlers) UserDetail() gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if h.users == nil {
+			h.renderMessage(c, http.StatusServiceUnavailable, "users", "Users", "User administration is not available.", "bad", "/_gm/")
+			return
+		}
+		userID, ok := parseUserIDParam(c)
+		if !ok {
+			return
+		}
+		account, err := h.users.GetAccount(c.Request.Context(), userID)
+		if err != nil {
+			if errors.Is(err, user.ErrAccountNotFound) {
+				h.renderMessage(c, http.StatusNotFound, "users", "User not found", "No such user, or the account has been soft-deleted.", "bad", "/_gm/users")
+				return
+			}
+			h.logger.Error("admin console: get user", zap.Error(err))
+			h.renderMessage(c, http.StatusInternalServerError, "users", "Users", "Failed to load the user.", "bad", "/_gm/users")
+			return
+		}
+		h.render(c, http.StatusOK, "user_detail", "users", account.Email, toUserDetailData(account))
+	}
+}
+
+// UserBlock handles POST /_gm/users/:user_id/block — applies a permanent block.
+func (h *AdminConsoleHandlers) UserBlock() gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if h.users == nil {
+			h.renderMessage(c, http.StatusServiceUnavailable, "users", "Users", "User administration is not available.", "bad", "/_gm/")
+			return
+		}
+		userID, ok := parseUserIDParam(c)
+		if !ok {
+			return
+		}
+		back := "/_gm/users/" + userID.String()
+		reason := strings.TrimSpace(c.PostForm("reason_code"))
+		if reason == "" {
+			h.renderMessage(c, http.StatusBadRequest, "users", "Invalid input", "A reason is required to block a user.", "bad", back)
+			return
+		}
+		_, err := h.users.ApplySanction(c.Request.Context(), user.ApplySanctionInput{
+			UserID:       userID,
+			SanctionCode: user.SanctionCodePermanentBlock,
+			Scope:        "account",
+			ReasonCode:   reason,
+			Actor:        actorFromContext(c),
+		})
+		if err != nil {
+			h.logger.Error("admin console: block user", zap.Error(err))
+			h.renderMessage(c, http.StatusInternalServerError, "users", "Block failed", "Failed to block the user.", "bad", back)
+			return
+		}
+		c.Redirect(http.StatusSeeOther, back)
+	}
+}
+
+// UserEntitlement handles POST /_gm/users/:user_id/entitlement.
+func (h *AdminConsoleHandlers) UserEntitlement() gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if h.users == nil {
+			h.renderMessage(c, http.StatusServiceUnavailable, "users", "Users", "User administration is not available.", "bad", "/_gm/")
+			return
+		}
+		userID, ok := parseUserIDParam(c)
+		if !ok {
+			return
+		}
+		back := "/_gm/users/" + userID.String()
+		tier := strings.TrimSpace(c.PostForm("tier"))
+		source := strings.TrimSpace(c.PostForm("source"))
+		if source == "" {
+			source = "admin"
+		}
+		_, err := h.users.ApplyEntitlement(c.Request.Context(), user.ApplyEntitlementInput{
+			UserID:     userID,
+			Tier:       tier,
+			Source:     source,
+			Actor:      actorFromContext(c),
+			ReasonCode: strings.TrimSpace(c.PostForm("reason_code")),
+		})
+		if err != nil {
+			if errors.Is(err, user.ErrInvalidInput) {
+				h.renderMessage(c, http.StatusBadRequest, "users", "Invalid input", "The entitlement request was rejected: check the tier.", "bad", back)
+				return
+			}
+			h.logger.Error("admin console: apply entitlement", zap.Error(err))
+			h.renderMessage(c, http.StatusInternalServerError, "users", "Entitlement failed", "Failed to update the entitlement.", "bad", back)
+			return
+		}
+		c.Redirect(http.StatusSeeOther, back)
+	}
+}
+
+// UserSoftDelete handles POST /_gm/users/:user_id/soft-delete.
+func (h *AdminConsoleHandlers) UserSoftDelete() gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if h.users == nil {
+			h.renderMessage(c, http.StatusServiceUnavailable, "users", "Users", "User administration is not available.", "bad", "/_gm/")
+			return
+		}
+		userID, ok := parseUserIDParam(c)
+		if !ok {
+			return
+		}
+		if err := h.users.SoftDelete(c.Request.Context(), userID, actorFromContext(c)); err != nil {
+			if errors.Is(err, user.ErrAccountNotFound) {
+				h.renderMessage(c, http.StatusNotFound, "users", "User not found", "No such user.", "bad", "/_gm/users")
+				return
+			}
+			// A cascade error does not undo the soft delete; log and proceed.
+			h.logger.Warn("admin console: soft-delete cascade returned error", zap.Error(err))
+		}
+		c.Redirect(http.StatusSeeOther, "/_gm/users")
+	}
+}
+
+// actorFromContext builds the admin ActorRef for audit trails from the
+// authenticated operator username stored by the Basic Auth middleware.
+func actorFromContext(c *gin.Context) user.ActorRef {
+	username, _ := basicauth.UsernameFromContext(c.Request.Context())
+	return user.ActorRef{Type: "admin", ID: username}
+}
+
+// toUsersListData maps an account page into the users list view model.
+func toUsersListData(page user.AccountPage) adminconsole.UsersListData {
+	data := adminconsole.UsersListData{
+		Items:    make([]adminconsole.UserRow, 0, len(page.Items)),
+		Page:     page.Page,
+		PageSize: page.PageSize,
+		Total:    page.Total,
+		PrevPage: page.Page - 1,
+		NextPage: page.Page + 1,
+		HasPrev:  page.Page > 1,
+		HasNext:  page.Page*page.PageSize < page.Total,
+	}
+	for _, account := range page.Items {
+		data.Items = append(data.Items, adminconsole.UserRow{
+			UserID:      account.UserID.String(),
+			Email:       account.Email,
+			UserName:    account.UserName,
+			DisplayName: account.DisplayName,
+			Tier:        account.Entitlement.Tier,
+			Blocked:     account.PermanentBlock,
+			Deleted:     account.DeletedAt != nil,
+			CreatedAt:   fmtConsoleTime(account.CreatedAt),
+		})
+	}
+	return data
+}
+
+// toUserDetailData maps an account aggregate into the detail view model.
+func toUserDetailData(account user.Account) adminconsole.UserDetailData {
+	data := adminconsole.UserDetailData{
+		UserID:            account.UserID.String(),
+		Email:             account.Email,
+		UserName:          account.UserName,
+		DisplayName:       account.DisplayName,
+		PreferredLanguage: account.PreferredLanguage,
+		TimeZone:          account.TimeZone,
+		DeclaredCountry:   account.DeclaredCountry,
+		Blocked:           account.PermanentBlock,
+		Deleted:           account.DeletedAt != nil,
+		CreatedAt:         fmtConsoleTime(account.CreatedAt),
+		UpdatedAt:         fmtConsoleTime(account.UpdatedAt),
+		Tier:              account.Entitlement.Tier,
+		IsPaid:            account.Entitlement.IsPaid,
+		EntitlementSource: account.Entitlement.Source,
+		EntitlementReason: account.Entitlement.ReasonCode,
+		EntitlementEnds:   fmtConsoleTimePtr(account.Entitlement.EndsAt),
+		Tiers:             consoleTiers,
+	}
+	for _, sanction := range account.ActiveSanctions {
+		data.Sanctions = append(data.Sanctions, adminconsole.SanctionView{
+			SanctionCode: sanction.SanctionCode,
+			Scope:        sanction.Scope,
+			ReasonCode:   sanction.ReasonCode,
+			AppliedAt:    fmtConsoleTime(sanction.AppliedAt),
+			ExpiresAt:    fmtConsoleTimePtr(sanction.ExpiresAt),
+		})
+	}
+	return data
+}
+
+// fmtConsoleTime renders a timestamp for display in the console.
+func fmtConsoleTime(t time.Time) string {
+	if t.IsZero() {
+		return ""
+	}
+	return t.UTC().Format("2006-01-02 15:04 UTC")
+}
+
+// fmtConsoleTimePtr renders an optional timestamp, returning "" when nil.
+func fmtConsoleTimePtr(t *time.Time) string {
+	if t == nil {
+		return ""
+	}
+	return fmtConsoleTime(*t)
+}

backend/internal/server/handlers_admin_console_users_test.go

@@ -0,0 +1,288 @@
+package server
+
+import (
+	"context"
+	"net/http"
+	"net/http/httptest"
+	"strings"
+	"testing"
+
+	"galaxy/backend/internal/adminconsole"
+	"galaxy/backend/internal/server/middleware/basicauth"
+	"galaxy/backend/internal/user"
+
+	"github.com/google/uuid"
+	"go.uber.org/zap"
+)
+
+// fakeUserAdmin records calls so the console handlers can be exercised without
+// a database.
+type fakeUserAdmin struct {
+	page    user.AccountPage
+	account user.Account
+	getErr  error
+
+	sanctionCalls   int
+	lastSanction    user.ApplySanctionInput
+	entitlementCall int
+	lastEntitlement user.ApplyEntitlementInput
+	softDeleteCalls int
+	lastSoftActor   user.ActorRef
+}
+
+func (f *fakeUserAdmin) ListAccounts(context.Context, int, int) (user.AccountPage, error) {
+	return f.page, nil
+}
+
+func (f *fakeUserAdmin) GetAccount(context.Context, uuid.UUID) (user.Account, error) {
+	return f.account, f.getErr
+}
+
+func (f *fakeUserAdmin) ApplySanction(_ context.Context, in user.ApplySanctionInput) (user.Account, error) {
+	f.sanctionCalls++
+	f.lastSanction = in
+	return f.account, nil
+}
+
+func (f *fakeUserAdmin) ApplyEntitlement(_ context.Context, in user.ApplyEntitlementInput) (user.Account, error) {
+	f.entitlementCall++
+	f.lastEntitlement = in
+	return f.account, nil
+}
+
+func (f *fakeUserAdmin) SoftDelete(_ context.Context, _ uuid.UUID, actor user.ActorRef) error {
+	f.softDeleteCalls++
+	f.lastSoftActor = actor
+	return nil
+}
+
+func newUsersConsoleRouter(t *testing.T, users UserAdmin) (http.Handler, *adminconsole.CSRF) {
+	t.Helper()
+	csrf := adminconsole.NewCSRF([]byte("test-key"))
+	handler, err := NewRouter(RouterDependencies{
+		Logger:        zap.NewNop(),
+		AdminVerifier: basicauth.NewStaticVerifier("secret"),
+		AdminConsole:  NewAdminConsoleHandlers(AdminConsoleDeps{CSRF: csrf, Users: users}),
+	})
+	if err != nil {
+		t.Fatalf("NewRouter: %v", err)
+	}
+	return handler, csrf
+}
+
+func TestConsoleUsersList(t *testing.T) {
+	fake := &fakeUserAdmin{page: user.AccountPage{
+		Items: []user.Account{
+			{UserID: uuid.New(), Email: "alice@example.test", UserName: "alice"},
+			{UserID: uuid.New(), Email: "bob@example.test", UserName: "bob", PermanentBlock: true},
+		},
+		Page: 1, PageSize: 50, Total: 2,
+	}}
+	router, _ := newUsersConsoleRouter(t, fake)
+
+	req := httptest.NewRequest(http.MethodGet, "/_gm/users", nil)
+	req.SetBasicAuth("ops", "secret")
+	rec := httptest.NewRecorder()
+	router.ServeHTTP(rec, req)
+
+	if rec.Code != http.StatusOK {
+		t.Fatalf("status = %d, want 200; body=%s", rec.Code, rec.Body.String())
+	}
+	body := rec.Body.String()
+	for _, want := range []string{"alice@example.test", "bob@example.test", "blocked", "page 1"} {
+		if !strings.Contains(body, want) {
+			t.Errorf("users list missing %q", want)
+		}
+	}
+}
+
+func TestConsoleUserDetailRendersForms(t *testing.T) {
+	id := uuid.New()
+	fake := &fakeUserAdmin{account: user.Account{
+		UserID: id, Email: "alice@example.test", UserName: "alice",
+		Entitlement: user.EntitlementSnapshot{Tier: user.TierFree},
+	}}
+	router, csrf := newUsersConsoleRouter(t, fake)
+
+	req := httptest.NewRequest(http.MethodGet, "/_gm/users/"+id.String(), nil)
+	req.SetBasicAuth("ops", "secret")
+	rec := httptest.NewRecorder()
+	router.ServeHTTP(rec, req)
+
+	if rec.Code != http.StatusOK {
+		t.Fatalf("status = %d, want 200; body=%s", rec.Code, rec.Body.String())
+	}
+	body := rec.Body.String()
+	for _, want := range []string{
+		"alice@example.test",
+		"Permanently block",
+		"Update entitlement",
+		"Soft-delete account",
+		csrf.Token("ops"),
+	} {
+		if !strings.Contains(body, want) {
+			t.Errorf("user detail missing %q", want)
+		}
+	}
+}
+
+func TestConsoleUserDetailNotFound(t *testing.T) {
+	fake := &fakeUserAdmin{getErr: user.ErrAccountNotFound}
+	router, _ := newUsersConsoleRouter(t, fake)
+
+	req := httptest.NewRequest(http.MethodGet, "/_gm/users/"+uuid.New().String(), nil)
+	req.SetBasicAuth("ops", "secret")
+	rec := httptest.NewRecorder()
+	router.ServeHTTP(rec, req)
+
+	if rec.Code != http.StatusNotFound {
+		t.Fatalf("status = %d, want 404", rec.Code)
+	}
+	if !strings.Contains(rec.Body.String(), "not found") {
+		t.Error("expected a not-found message")
+	}
+}
+
+func TestConsoleUserBlock(t *testing.T) {
+	id := uuid.New()
+	fake := &fakeUserAdmin{account: user.Account{UserID: id}}
+	router, csrf := newUsersConsoleRouter(t, fake)
+
+	form := "_csrf=" + csrf.Token("ops") + "&reason_code=spam"
+	req := httptest.NewRequest(http.MethodPost, "http://galaxy.lan/_gm/users/"+id.String()+"/block", strings.NewReader(form))
+	req.Header.Set("Content-Type", "application/x-www-form-urlencoded")
+	req.Header.Set("Origin", "https://galaxy.lan")
+	req.SetBasicAuth("ops", "secret")
+	rec := httptest.NewRecorder()
+	router.ServeHTTP(rec, req)
+
+	if rec.Code != http.StatusSeeOther {
+		t.Fatalf("status = %d, want 303; body=%s", rec.Code, rec.Body.String())
+	}
+	if fake.sanctionCalls != 1 {
+		t.Fatalf("ApplySanction called %d times, want 1", fake.sanctionCalls)
+	}
+	if fake.lastSanction.SanctionCode != user.SanctionCodePermanentBlock {
+		t.Errorf("sanction code = %q, want permanent_block", fake.lastSanction.SanctionCode)
+	}
+	if fake.lastSanction.Scope != "account" {
+		t.Errorf("scope = %q, want account", fake.lastSanction.Scope)
+	}
+	if fake.lastSanction.ReasonCode != "spam" {
+		t.Errorf("reason = %q, want spam", fake.lastSanction.ReasonCode)
+	}
+	if fake.lastSanction.Actor.Type != "admin" || fake.lastSanction.Actor.ID != "ops" {
+		t.Errorf("actor = %+v, want admin/ops", fake.lastSanction.Actor)
+	}
+	if fake.lastSanction.UserID != id {
+		t.Errorf("sanction user id = %s, want %s", fake.lastSanction.UserID, id)
+	}
+}
+
+func TestConsoleUserBlockMissingReason(t *testing.T) {
+	id := uuid.New()
+	fake := &fakeUserAdmin{account: user.Account{UserID: id}}
+	router, csrf := newUsersConsoleRouter(t, fake)
+
+	form := "_csrf=" + csrf.Token("ops")
+	req := httptest.NewRequest(http.MethodPost, "http://galaxy.lan/_gm/users/"+id.String()+"/block", strings.NewReader(form))
+	req.Header.Set("Content-Type", "application/x-www-form-urlencoded")
+	req.Header.Set("Origin", "https://galaxy.lan")
+	req.SetBasicAuth("ops", "secret")
+	rec := httptest.NewRecorder()
+	router.ServeHTTP(rec, req)
+
+	if rec.Code != http.StatusBadRequest {
+		t.Fatalf("status = %d, want 400", rec.Code)
+	}
+	if fake.sanctionCalls != 0 {
+		t.Errorf("ApplySanction must not be called without a reason")
+	}
+}
+
+func TestConsoleUserBlockRejectsBadCSRF(t *testing.T) {
+	id := uuid.New()
+	fake := &fakeUserAdmin{account: user.Account{UserID: id}}
+	router, _ := newUsersConsoleRouter(t, fake)
+
+	req := httptest.NewRequest(http.MethodPost, "http://galaxy.lan/_gm/users/"+id.String()+"/block", strings.NewReader("reason_code=spam"))
+	req.Header.Set("Content-Type", "application/x-www-form-urlencoded")
+	req.Header.Set("Origin", "https://galaxy.lan")
+	req.SetBasicAuth("ops", "secret")
+	rec := httptest.NewRecorder()
+	router.ServeHTTP(rec, req)
+
+	if rec.Code != http.StatusForbidden {
+		t.Fatalf("status = %d, want 403", rec.Code)
+	}
+	if fake.sanctionCalls != 0 {
+		t.Errorf("ApplySanction must not run when the CSRF token is missing")
+	}
+}
+
+func TestConsoleUserEntitlement(t *testing.T) {
+	id := uuid.New()
+	fake := &fakeUserAdmin{account: user.Account{UserID: id}}
+	router, csrf := newUsersConsoleRouter(t, fake)
+
+	form := "_csrf=" + csrf.Token("ops") + "&tier=monthly&source=admin&reason_code=promo"
+	req := httptest.NewRequest(http.MethodPost, "http://galaxy.lan/_gm/users/"+id.String()+"/entitlement", strings.NewReader(form))
+	req.Header.Set("Content-Type", "application/x-www-form-urlencoded")
+	req.Header.Set("Origin", "https://galaxy.lan")
+	req.SetBasicAuth("ops", "secret")
+	rec := httptest.NewRecorder()
+	router.ServeHTTP(rec, req)
+
+	if rec.Code != http.StatusSeeOther {
+		t.Fatalf("status = %d, want 303; body=%s", rec.Code, rec.Body.String())
+	}
+	if fake.entitlementCall != 1 {
+		t.Fatalf("ApplyEntitlement called %d times, want 1", fake.entitlementCall)
+	}
+	if fake.lastEntitlement.Tier != user.TierMonthly {
+		t.Errorf("tier = %q, want monthly", fake.lastEntitlement.Tier)
+	}
+	if fake.lastEntitlement.Actor.ID != "ops" {
+		t.Errorf("actor id = %q, want ops", fake.lastEntitlement.Actor.ID)
+	}
+}
+
+func TestConsoleUserSoftDelete(t *testing.T) {
+	id := uuid.New()
+	fake := &fakeUserAdmin{account: user.Account{UserID: id}}
+	router, csrf := newUsersConsoleRouter(t, fake)
+
+	form := "_csrf=" + csrf.Token("ops")
+	req := httptest.NewRequest(http.MethodPost, "http://galaxy.lan/_gm/users/"+id.String()+"/soft-delete", strings.NewReader(form))
+	req.Header.Set("Content-Type", "application/x-www-form-urlencoded")
+	req.Header.Set("Origin", "https://galaxy.lan")
+	req.SetBasicAuth("ops", "secret")
+	rec := httptest.NewRecorder()
+	router.ServeHTTP(rec, req)
+
+	if rec.Code != http.StatusSeeOther {
+		t.Fatalf("status = %d, want 303", rec.Code)
+	}
+	if got := rec.Header().Get("Location"); got != "/_gm/users" {
+		t.Errorf("redirect Location = %q, want /_gm/users", got)
+	}
+	if fake.softDeleteCalls != 1 {
+		t.Fatalf("SoftDelete called %d times, want 1", fake.softDeleteCalls)
+	}
+	if fake.lastSoftActor.ID != "ops" {
+		t.Errorf("soft-delete actor = %q, want ops", fake.lastSoftActor.ID)
+	}
+}
+
+func TestConsoleUsersUnavailable(t *testing.T) {
+	router, _ := newUsersConsoleRouter(t, nil) // no user service wired
+
+	req := httptest.NewRequest(http.MethodGet, "/_gm/users", nil)
+	req.SetBasicAuth("ops", "secret")
+	rec := httptest.NewRecorder()
+	router.ServeHTTP(rec, req)
+
+	if rec.Code != http.StatusServiceUnavailable {
+		t.Fatalf("status = %d, want 503", rec.Code)
+	}
+}

backend/internal/server/router.go

@@ -81,6 +81,13 @@ type RouterDependencies struct {
 	AdminGeo *AdminGeoHandlers
 	InternalSessions *InternalSessionsHandlers
 	InternalUsers *InternalUsersHandlers
+
+	// AdminConsole, when non-nil, mounts the server-rendered operator
+	// console under the `/_gm` route group behind the same admin Basic
+	// Auth verifier as `/api/v1/admin`. A nil value leaves the console
+	// unmounted, which keeps routers built without console wiring (the
+	// contract test, most unit tests) unchanged.
+	AdminConsole *AdminConsoleHandlers
 }
 
 // NewRouter constructs the backend gin engine wired with the documented
@@ -123,6 +130,7 @@ func NewRouter(deps RouterDependencies) (http.Handler, error) {
 	registerUserRoutes(router, instruments, deps)
 	registerAdminRoutes(router, instruments, deps)
 	registerInternalRoutes(router, instruments, deps)
+	registerAdminConsoleRoutes(router, deps)
 
 	router.NoMethod(func(c *gin.Context) {
 		if allow := allowedMethodsForPath(c.Request.URL.Path); allow != "" {
@@ -364,6 +372,57 @@ func registerInternalRoutes(router *gin.Engine, instruments *metrics.Instruments
 	users.GET("/:user_id/account-internal", deps.InternalUsers.GetAccountInternal())
 }
 
+// registerAdminConsoleRoutes mounts the server-rendered operator console under
+// `/_gm` when deps.AdminConsole is wired. The group reuses the same admin Basic
+// Auth verifier as `/api/v1/admin`; the CSRF guard then protects every
+// state-changing request. A nil AdminConsole leaves the surface unmounted.
+func registerAdminConsoleRoutes(router *gin.Engine, deps RouterDependencies) {
+	if deps.AdminConsole == nil {
+		return
+	}
+
+	group := router.Group("/_gm")
+	group.Use(basicauth.Middleware(deps.AdminVerifier, adminBasicAuthRealm))
+	group.Use(deps.AdminConsole.RequireCSRF())
+
+	group.GET("/assets/*filepath", deps.AdminConsole.Asset())
+	group.GET("", deps.AdminConsole.Dashboard())
+	group.GET("/", deps.AdminConsole.Dashboard())
+
+	group.GET("/users", deps.AdminConsole.UsersList())
+	group.GET("/users/:user_id", deps.AdminConsole.UserDetail())
+	group.POST("/users/:user_id/block", deps.AdminConsole.UserBlock())
+	group.POST("/users/:user_id/entitlement", deps.AdminConsole.UserEntitlement())
+	group.POST("/users/:user_id/soft-delete", deps.AdminConsole.UserSoftDelete())
+
+	group.GET("/games", deps.AdminConsole.GamesList())
+	group.POST("/games", deps.AdminConsole.GameCreate())
+	group.GET("/games/:game_id", deps.AdminConsole.GameDetail())
+	group.POST("/games/:game_id/force-start", deps.AdminConsole.GameForceStart())
+	group.POST("/games/:game_id/force-stop", deps.AdminConsole.GameForceStop())
+	group.POST("/games/:game_id/ban-member", deps.AdminConsole.GameBanMember())
+	group.POST("/games/:game_id/runtime/restart", deps.AdminConsole.RuntimeRestart())
+	group.POST("/games/:game_id/runtime/patch", deps.AdminConsole.RuntimePatch())
+	group.POST("/games/:game_id/runtime/force-next-turn", deps.AdminConsole.RuntimeForceNextTurn())
+
+	group.GET("/engine-versions", deps.AdminConsole.EngineVersionsList())
+	group.POST("/engine-versions", deps.AdminConsole.EngineVersionRegister())
+	group.POST("/engine-versions/:version/disable", deps.AdminConsole.EngineVersionDisable())
+
+	group.GET("/operators", deps.AdminConsole.OperatorsList())
+	group.POST("/operators", deps.AdminConsole.OperatorCreate())
+	group.POST("/operators/:username/disable", deps.AdminConsole.OperatorDisable())
+	group.POST("/operators/:username/enable", deps.AdminConsole.OperatorEnable())
+	group.POST("/operators/:username/reset-password", deps.AdminConsole.OperatorResetPassword())
+
+	group.GET("/mail", deps.AdminConsole.MailPage())
+	group.GET("/mail/deliveries/:delivery_id", deps.AdminConsole.MailDeliveryDetail())
+	group.POST("/mail/deliveries/:delivery_id/resend", deps.AdminConsole.MailResend())
+	group.GET("/notifications", deps.AdminConsole.NotificationsPage())
+	group.GET("/broadcast", deps.AdminConsole.BroadcastForm())
+	group.POST("/broadcast", deps.AdminConsole.BroadcastSend())
+}
+
 // allowedMethodsForPath returns the comma-separated list of methods
 // the gin router accepts on requestPath. Only the probe paths declare
 // a non-empty list so NoMethod can advertise a useful `Allow` header

docs/ARCHITECTURE.md

@@ -581,6 +581,36 @@ directly.
   `/api/v1/admin/notifications/*`) reuse the per-domain logic of the
   module they target.
 
+### 14.1 Operator console (`/_gm`)
+
+`backend` also serves a server-rendered operator console under the `/_gm`
+route group — the human-facing surface for the admin operations otherwise
+exposed as JSON under `/api/v1/admin/*`. It reuses the `admin_accounts`
+Basic Auth verifier and renders pages with the standard library's
+`html/template` (navigation by path and query, Post/Redirect/Get on
+writes; no client framework or build step).
+
+Unlike the internal-only JSON admin API, the console is reachable from the
+public edge: Caddy routes `/_gm/*` to the gateway public listener, which
+classifies it as the `admin` anti-abuse class (per-IP rate limit, body and
+method limits) and reverse-proxies it to `backend`'s `/_gm` surface. The
+gateway preserves the inbound `Host` and relays the backend's 401 Basic
+Auth challenge unchanged, so the browser shows its native credential
+dialog. Authentication is enforced by `backend`; the gateway contributes
+only the edge anti-abuse layer.
+
+State-changing requests are guarded against CSRF by a stateless token
+(HMAC-SHA256 over the authenticated username, keyed by
+`BACKEND_ADMIN_CONSOLE_CSRF_KEY`; a per-process random key is used when the
+variable is unset) plus a same-origin `Origin`/`Referer` check.
+
+The console landing page is a dashboard that surfaces backend-visible
+operational signals — database reachability, per-status game-runtime counts,
+and mail/notification queue depths — read directly through the persistence
+layer; richer historical metrics come from the Prometheus exporters on
+`backend` and `gateway` (see [§17](#17-observability)). See
+`backend/docs/admin-console.md` for the console design.
+
 ## 15. Transport Security Model (gateway boundary)
 
 This section describes the secure exchange model between client and
@@ -823,7 +853,8 @@ business validation and authorisation.
 | Session revocation propagation                           | backend → gateway       | `session_invalidation` over the gRPC push stream flips the gateway-side cache entry to revoked and closes any active push stream. |
 | Authorisation, ownership, state transitions              | backend                 | `X-User-ID` is the sole identity input on the user surface.                                     |
 | Edge rate limiting                                       | gateway                 | Backend has no rate-limit responsibility in MVP.                                                |
-| Admin authentication                                     | backend                 | Basic Auth against `admin_accounts`.                                                            |
+| Admin authentication                                     | backend                 | Basic Auth against `admin_accounts`; the `/_gm` operator console reuses the same verifier.       |
+| Admin console CSRF                                        | backend                 | Stateless HMAC token (`BACKEND_ADMIN_CONSOLE_CSRF_KEY`) + same-origin `Origin`/`Referer` check on `/_gm` writes. |
 | Engine API authentication                                | network                 | Engine listens only on the trusted network; backend is the only caller.                         |
 
 ### Backend ↔ Gateway trust
@@ -857,6 +888,19 @@ addition.
 - Health probes are unauthenticated `GET /healthz` (process liveness) and
   `GET /readyz` (Postgres reachable, migrations applied, gRPC listener
   bound). Probes are excluded from anti-replay and rate limiting.
+- **Collection (dev, production mirror).** The long-lived dev environment
+  (`tools/dev-deploy/`) runs a full metrics + logs + traces stack on its
+  internal network with no host ports: Prometheus scrapes the backend
+  (`:9100`) and gateway (`:9191`) endpoints plus `node-exporter` and
+  cAdvisor; Tempo ingests OTLP traces from backend and gateway; Loki
+  stores container logs shipped by promtail (Docker service-discovery on
+  the `galaxy.stack=dev-deploy` label). Grafana (provisioned datasources
+  + dashboards) and the Mailpit capture UI are reached only through the
+  operator console's single `/_gm` Basic Auth gate (§14.1) — at
+  `/_gm/grafana/` and `/_gm/mailpit/` — so one password covers the
+  console and both UIs. Retention is tuned small (Prometheus 15d, Loki
+  7d, Tempo 3d). The same compose fragment is meant to back production.
+  See `tools/dev-deploy/monitoring/README.md`.
 
 ## 18. CI and Environments
 

docs/FUNCTIONAL.md

@@ -1162,6 +1162,31 @@ operator's password manager can match it across deployments.
 After the first deployment, the bootstrap password should be
 rotated through the admin surface.
 
+### 10.2.1 Operator console (`/_gm`)
+
+Administrators drive these operations either programmatically through
+the JSON admin API or through a server-rendered web console at `/_gm`.
+The console authenticates with the same Basic Auth credentials: opening
+any `/_gm` page prompts the browser's native credential dialog, and the
+operator stays signed in for the session. Navigation is by ordinary
+links and query parameters; every change is submitted as a form and
+answered with a redirect back to the affected page.
+
+The console is the only admin surface reachable from outside the trusted
+network. It is fronted by the gateway, so it inherits the same edge rate
+limiting and request limits as the public API, and it carries an
+anti-CSRF token on every change. The JSON admin API stays internal to
+the deployment.
+
+The console landing page is a dashboard that summarises operational
+health: whether the backend is ready and the database reachable, how many
+game runtimes sit in each state, and the depth of the mail and
+notification queues. It is a read-only point-in-time view for quick
+triage, not a metrics history. The console nav also links to Grafana
+(metrics, logs and traces) and the Mailpit capture UI, which the
+deployment serves under the same `/_gm` Basic Auth gate — one sign-in
+covers the console and both UIs.
+
 ### 10.3 Admin account management
 
 Existing admins can list other admins, create new ones, look up a

docs/FUNCTIONAL_ru.md

@@ -1197,6 +1197,31 @@ deployments.
 После первого деплоя bootstrap-пароль должен быть ротирован
 через admin-surface.
 
+### 10.2.1 Операторская консоль (`/_gm`)
+
+Администраторы выполняют эти операции либо программно через JSON
+admin-API, либо через серверно-рендеримую веб-консоль на `/_gm`.
+Консоль аутентифицируется теми же Basic Auth-учётными данными:
+открытие любой страницы `/_gm` вызывает нативный диалог браузера для
+ввода учётных данных, и оператор остаётся залогинен на время сессии.
+Навигация — обычными ссылками и query-параметрами; каждое изменение
+отправляется формой и завершается редиректом обратно на затронутую
+страницу.
+
+Консоль — единственная admin-поверхность, достижимая извне
+доверенной сети. Она проксируется через gateway, поэтому наследует те
+же edge-rate-limiting и лимиты запросов, что и публичный API, и несёт
+анти-CSRF-токен на каждом изменении. JSON admin-API остаётся
+внутренним для деплоя.
+
+Стартовая страница консоли — дашборд, сводящий операционное
+здоровье: готов ли backend и доступна ли БД, сколько игровых рантаймов
+в каждом состоянии, какова глубина очередей почты и уведомлений. Это
+read-only-срез на текущий момент для быстрой диагностики, не история
+метрик. Навигация консоли также ведёт в Grafana (метрики, логи и
+трейсы) и в UI захвата почты Mailpit, которые деплой отдаёт под тем же
+шлюзом Basic Auth `/_gm` — один вход покрывает консоль и оба UI.
+
 ### 10.3 Управление admin-аккаунтами
 
 Существующие админы могут перечислять других админов, создавать

gateway/README.md

@@ -178,6 +178,30 @@ bootstrap or asset traffic through a pluggable public handler or proxy.
 That traffic belongs to dedicated public route classes and must not share rate
 limit buckets or abuse counters with the public auth API.
 
+### Operator Console Proxy (`/_gm`)
+
+The gateway also fronts the backend operator console. The edge Caddy routes
+`/_gm` and `/_gm/*` to this public listener; the gateway classifies that
+traffic as the `admin` public route class and reverse-proxies it to the
+backend at `GATEWAY_BACKEND_HTTP_URL`, preserving the request path and the
+inbound `Host` header (so the backend's same-origin CSRF check observes the
+public host).
+
+Authentication is delegated entirely to the backend (HTTP Basic Auth against
+`admin_accounts`): the backend's `401` challenge is relayed unchanged so the
+browser shows its native credential dialog. The gateway contributes only the
+edge anti-abuse layer — a per-IP rate limit, a body size limit, and a
+`GET`/`HEAD`/`POST` method allow-list for the class — and answers
+`502 bad_gateway` when the backend is unreachable.
+
+The `admin` class carries its own budgets, isolated from the other public
+classes:
+
+- `GATEWAY_PUBLIC_HTTP_ANTI_ABUSE_ADMIN_MAX_BODY_BYTES` (default `65536`);
+- `GATEWAY_PUBLIC_HTTP_ANTI_ABUSE_ADMIN_RATE_LIMIT_REQUESTS` (default `120`);
+- `GATEWAY_PUBLIC_HTTP_ANTI_ABUSE_ADMIN_RATE_LIMIT_WINDOW` (default `1m`);
+- `GATEWAY_PUBLIC_HTTP_ANTI_ABUSE_ADMIN_RATE_LIMIT_BURST` (default `40`).
+
 ### Operational Admin Surface
 
 The gateway may expose one private operational HTTP listener used for metrics.

gateway/cmd/gateway/main.go

@@ -78,6 +78,15 @@ func run(ctx context.Context) (err error) {
 		AuthService: authServiceAdapter{rest: backend.REST()},
 	}
 
+	adminConsoleProxy, err := restapi.NewBackendConsoleProxy(cfg.Backend.HTTPBaseURL, logger)
+	if err != nil {
+		_ = backend.Close()
+		_ = telemetryRuntime.Shutdown(context.Background())
+		_ = logging.Sync(logger)
+		return fmt.Errorf("build admin console proxy: %w", err)
+	}
+	publicRESTDeps.AdminConsoleProxy = adminConsoleProxy
+
 	grpcDeps, components, cleanup, err := newAuthenticatedGRPCDependencies(ctx, cfg, logger, telemetryRuntime, backend)
 	if err != nil {
 		_ = backend.Close()

gateway/internal/config/config.go

@@ -276,6 +276,22 @@ const (
 	// configures the public_misc rate-limit burst.
 	publicMiscRateLimitBurstEnvVar = "GATEWAY_PUBLIC_HTTP_ANTI_ABUSE_PUBLIC_MISC_RATE_LIMIT_BURST"
 
+	// adminMaxBodyBytesEnvVar names the environment variable that configures
+	// the maximum accepted request body size for the admin console class.
+	adminMaxBodyBytesEnvVar = "GATEWAY_PUBLIC_HTTP_ANTI_ABUSE_ADMIN_MAX_BODY_BYTES"
+
+	// adminRateLimitRequestsEnvVar names the environment variable that
+	// configures the admin console request budget per window.
+	adminRateLimitRequestsEnvVar = "GATEWAY_PUBLIC_HTTP_ANTI_ABUSE_ADMIN_RATE_LIMIT_REQUESTS"
+
+	// adminRateLimitWindowEnvVar names the environment variable that configures
+	// the admin console rate-limit window.
+	adminRateLimitWindowEnvVar = "GATEWAY_PUBLIC_HTTP_ANTI_ABUSE_ADMIN_RATE_LIMIT_WINDOW"
+
+	// adminRateLimitBurstEnvVar names the environment variable that configures
+	// the admin console rate-limit burst.
+	adminRateLimitBurstEnvVar = "GATEWAY_PUBLIC_HTTP_ANTI_ABUSE_ADMIN_RATE_LIMIT_BURST"
+
 	// sendEmailCodeIdentityRateLimitRequestsEnvVar names the environment
 	// variable that configures the send-email-code identity request budget per
 	// window.
@@ -372,6 +388,14 @@ const (
 	defaultPublicMiscRateLimitRequests = 30
 	defaultPublicMiscRateLimitBurst = 10
 
+	// Admin console class: sized for a human operator clicking through pages
+	// and submitting forms, while still throttling Basic Auth brute-force at
+	// the edge. The body budget accommodates form posts.
+	defaultAdminMaxBodyBytes = int64(65536)
+
+	defaultAdminRateLimitRequests = 120
+	defaultAdminRateLimitBurst = 40
+
 	defaultSendEmailCodeIdentityRateLimitRequests = 3
 	defaultSendEmailCodeIdentityRateLimitBurst = 1
 
@@ -439,6 +463,11 @@ type PublicHTTPAntiAbuseConfig struct {
 	// PublicMisc applies to the stable public_misc route class.
 	PublicMisc PublicRoutePolicyConfig
 
+	// Admin applies to the stable admin route class — the `/_gm` operator
+	// console reverse-proxied to the backend. Only per-IP limiting applies;
+	// the class carries no identity buckets.
+	Admin PublicRoutePolicyConfig
+
 	// SendEmailCodeIdentity applies the additional identity limiter for
 	// send-email-code.
 	SendEmailCodeIdentity PublicAuthIdentityPolicyConfig
@@ -708,6 +737,14 @@ func DefaultPublicHTTPConfig() PublicHTTPConfig {
 					Burst: defaultPublicMiscRateLimitBurst,
 				},
 			},
+			Admin: PublicRoutePolicyConfig{
+				MaxBodyBytes: defaultAdminMaxBodyBytes,
+				RateLimit: PublicRateLimitConfig{
+					Requests: defaultAdminRateLimitRequests,
+					Window: defaultClassRateLimitWindow,
+					Burst: defaultAdminRateLimitBurst,
+				},
+			},
 			SendEmailCodeIdentity: PublicAuthIdentityPolicyConfig{
 				RateLimit: PublicRateLimitConfig{
 					Requests: defaultSendEmailCodeIdentityRateLimitRequests,
@@ -1092,6 +1129,18 @@ func LoadFromEnv() (Config, error) {
 	}
 	cfg.PublicHTTP.AntiAbuse.PublicMisc = publicMiscPolicy
 
+	adminPolicy, err := loadPublicRoutePolicyConfigFromEnv(
+		cfg.PublicHTTP.AntiAbuse.Admin,
+		adminMaxBodyBytesEnvVar,
+		adminRateLimitRequestsEnvVar,
+		adminRateLimitWindowEnvVar,
+		adminRateLimitBurstEnvVar,
+	)
+	if err != nil {
+		return Config{}, err
+	}
+	cfg.PublicHTTP.AntiAbuse.Admin = adminPolicy
+
 	sendIdentityPolicy, err := loadPublicAuthIdentityPolicyConfigFromEnv(
 		cfg.PublicHTTP.AntiAbuse.SendEmailCodeIdentity,
 		sendEmailCodeIdentityRateLimitRequestsEnvVar,
@@ -1247,6 +1296,9 @@ func LoadFromEnv() (Config, error) {
 	if err := validatePublicRoutePolicyConfig(cfg.PublicHTTP.AntiAbuse.PublicMisc, publicMiscMaxBodyBytesEnvVar, publicMiscRateLimitRequestsEnvVar, publicMiscRateLimitWindowEnvVar, publicMiscRateLimitBurstEnvVar); err != nil {
 		return Config{}, err
 	}
+	if err := validatePublicRoutePolicyConfig(cfg.PublicHTTP.AntiAbuse.Admin, adminMaxBodyBytesEnvVar, adminRateLimitRequestsEnvVar, adminRateLimitWindowEnvVar, adminRateLimitBurstEnvVar); err != nil {
+		return Config{}, err
+	}
 	if err := validatePublicAuthIdentityPolicyConfig(cfg.PublicHTTP.AntiAbuse.SendEmailCodeIdentity, sendEmailCodeIdentityRateLimitRequestsEnvVar, sendEmailCodeIdentityRateLimitWindowEnvVar, sendEmailCodeIdentityRateLimitBurstEnvVar); err != nil {
 		return Config{}, err
 	}

gateway/internal/restapi/admin_proxy.go

@@ -0,0 +1,49 @@
+package restapi
+
+import (
+	"fmt"
+	"net/http"
+	"net/http/httputil"
+	"net/url"
+
+	"go.uber.org/zap"
+)
+
+// NewBackendConsoleProxy builds the reverse proxy that forwards operator
+// console traffic (`/_gm` and `/_gm/*`) to the backend at backendBaseURL.
+//
+// The proxy is intentionally thin: it preserves the inbound request path and
+// the inbound Host header — the latter so the backend's same-origin CSRF check
+// observes the public host rather than the internal upstream — and relays the
+// backend response unchanged, including its 401 Basic Auth challenge. It
+// answers 502 when the backend is unreachable. Authentication, rendering, and
+// every state change live in the backend; the gateway contributes only the
+// public anti-abuse layer that runs ahead of this handler.
+func NewBackendConsoleProxy(backendBaseURL string, logger *zap.Logger) (http.Handler, error) {
+	target, err := url.Parse(backendBaseURL)
+	if err != nil {
+		return nil, fmt.Errorf("parse backend base URL %q: %w", backendBaseURL, err)
+	}
+	if target.Scheme == "" || target.Host == "" {
+		return nil, fmt.Errorf("backend base URL %q must be absolute", backendBaseURL)
+	}
+
+	if logger == nil {
+		logger = zap.NewNop()
+	}
+	logger = logger.Named("admin_console_proxy")
+
+	return &httputil.ReverseProxy{
+		Rewrite: func(pr *httputil.ProxyRequest) {
+			pr.SetURL(target)
+			// SetURL clears Out.Host so the target host is used; restore the
+			// inbound Host so the backend sees the public origin.
+			pr.Out.Host = pr.In.Host
+		},
+		ErrorHandler: func(w http.ResponseWriter, r *http.Request, err error) {
+			logger.Warn("admin console upstream error",
+				zap.String("path", r.URL.Path), zap.Error(err))
+			w.WriteHeader(http.StatusBadGateway)
+		},
+	}, nil
+}

gateway/internal/restapi/admin_proxy_test.go

@@ -0,0 +1,198 @@
+package restapi
+
+import (
+	"context"
+	"io"
+	"net/http"
+	"net/http/httptest"
+	"strings"
+	"sync/atomic"
+	"testing"
+	"time"
+
+	"galaxy/gateway/internal/config"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+// proxyRequest builds a test request whose context carries a cancellation
+// signal. A real http.Server always supplies one; httptest.NewRequest does not,
+// and without it httputil.ReverseProxy falls back to the legacy CloseNotifier
+// path, which panics under gin's ResponseWriter wrapping an
+// httptest.ResponseRecorder. Cancelling at test cleanup keeps the context live
+// for the synchronous ServeHTTP call.
+func proxyRequest(t *testing.T, method, target string, body io.Reader) *http.Request {
+	t.Helper()
+	req := httptest.NewRequest(method, target, body)
+	ctx, cancel := context.WithCancel(req.Context())
+	t.Cleanup(cancel)
+	return req.WithContext(ctx)
+}
+
+func TestAdminConsoleProxyForwardsToBackend(t *testing.T) {
+	var gotPath, gotHost, gotAuth string
+	backend := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		gotPath = r.URL.Path
+		gotHost = r.Host
+		gotAuth = r.Header.Get("Authorization")
+		w.Header().Set("Content-Type", "text/html; charset=utf-8")
+		_, _ = w.Write([]byte("<h1>Dashboard</h1>"))
+	}))
+	defer backend.Close()
+
+	proxy, err := NewBackendConsoleProxy(backend.URL, nil)
+	require.NoError(t, err)
+	handler := newPublicHandlerWithConfig(config.DefaultPublicHTTPConfig(), ServerDependencies{AdminConsoleProxy: proxy})
+
+	req := proxyRequest(t, http.MethodGet, "http://galaxy.lan/_gm/", nil)
+	req.SetBasicAuth("ops", "secret")
+	rec := httptest.NewRecorder()
+	handler.ServeHTTP(rec, req)
+
+	require.Equal(t, http.StatusOK, rec.Code)
+	assert.Contains(t, rec.Body.String(), "Dashboard")
+	assert.Equal(t, "/_gm/", gotPath)
+	assert.Equal(t, "galaxy.lan", gotHost, "inbound Host must be preserved for same-origin CSRF checks")
+	assert.True(t, strings.HasPrefix(gotAuth, "Basic "), "Authorization header must be forwarded to the backend")
+}
+
+func TestAdminConsoleProxyForwardsFormPost(t *testing.T) {
+	var gotPath, gotBody, gotContentType string
+	backend := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		gotPath = r.URL.Path
+		gotContentType = r.Header.Get("Content-Type")
+		body, _ := io.ReadAll(r.Body)
+		gotBody = string(body)
+		w.WriteHeader(http.StatusSeeOther)
+	}))
+	defer backend.Close()
+
+	proxy, err := NewBackendConsoleProxy(backend.URL, nil)
+	require.NoError(t, err)
+	handler := newPublicHandlerWithConfig(config.DefaultPublicHTTPConfig(), ServerDependencies{AdminConsoleProxy: proxy})
+
+	const form = "_csrf=token&reason=spam"
+	req := proxyRequest(t, http.MethodPost, "http://galaxy.lan/_gm/users/1/sanctions", strings.NewReader(form))
+	req.Header.Set("Content-Type", "application/x-www-form-urlencoded")
+	req.SetBasicAuth("ops", "secret")
+	rec := httptest.NewRecorder()
+	handler.ServeHTTP(rec, req)
+
+	require.Equal(t, http.StatusSeeOther, rec.Code)
+	assert.Equal(t, "/_gm/users/1/sanctions", gotPath)
+	assert.Equal(t, form, gotBody, "request body must reach the backend intact through the anti-abuse buffer")
+	assert.Contains(t, gotContentType, "x-www-form-urlencoded")
+}
+
+func TestAdminConsoleProxyRelaysAuthChallenge(t *testing.T) {
+	backend := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, _ *http.Request) {
+		w.Header().Set("WWW-Authenticate", `Basic realm="galaxy-admin"`)
+		w.WriteHeader(http.StatusUnauthorized)
+	}))
+	defer backend.Close()
+
+	proxy, err := NewBackendConsoleProxy(backend.URL, nil)
+	require.NoError(t, err)
+	handler := newPublicHandlerWithConfig(config.DefaultPublicHTTPConfig(), ServerDependencies{AdminConsoleProxy: proxy})
+
+	req := proxyRequest(t, http.MethodGet, "http://galaxy.lan/_gm/", nil)
+	rec := httptest.NewRecorder()
+	handler.ServeHTTP(rec, req)
+
+	require.Equal(t, http.StatusUnauthorized, rec.Code)
+	assert.Contains(t, rec.Header().Get("WWW-Authenticate"), "Basic")
+}
+
+func TestAdminConsoleProxyRejectsDisallowedMethod(t *testing.T) {
+	var hits int32
+	backend := httptest.NewServer(http.HandlerFunc(func(http.ResponseWriter, *http.Request) {
+		atomic.AddInt32(&hits, 1)
+	}))
+	defer backend.Close()
+
+	proxy, err := NewBackendConsoleProxy(backend.URL, nil)
+	require.NoError(t, err)
+	handler := newPublicHandlerWithConfig(config.DefaultPublicHTTPConfig(), ServerDependencies{AdminConsoleProxy: proxy})
+
+	req := proxyRequest(t, http.MethodDelete, "http://galaxy.lan/_gm/users/1", nil)
+	rec := httptest.NewRecorder()
+	handler.ServeHTTP(rec, req)
+
+	assert.Equal(t, http.StatusMethodNotAllowed, rec.Code)
+	assert.Equal(t, int32(0), atomic.LoadInt32(&hits), "backend must not be reached for a rejected method")
+}
+
+func TestAdminConsoleProxyRejectsOversizedBody(t *testing.T) {
+	backend := httptest.NewServer(http.HandlerFunc(func(http.ResponseWriter, *http.Request) {}))
+	defer backend.Close()
+
+	proxy, err := NewBackendConsoleProxy(backend.URL, nil)
+	require.NoError(t, err)
+	cfg := config.DefaultPublicHTTPConfig()
+	cfg.AntiAbuse.Admin.MaxBodyBytes = 8
+	handler := newPublicHandlerWithConfig(cfg, ServerDependencies{AdminConsoleProxy: proxy})
+
+	req := proxyRequest(t, http.MethodPost, "http://galaxy.lan/_gm/users/1/sanctions",
+		strings.NewReader("this body is well beyond eight bytes"))
+	req.Header.Set("Content-Type", "application/x-www-form-urlencoded")
+	rec := httptest.NewRecorder()
+	handler.ServeHTTP(rec, req)
+
+	assert.Equal(t, http.StatusRequestEntityTooLarge, rec.Code)
+}
+
+func TestAdminConsoleProxyRateLimitsPerIP(t *testing.T) {
+	backend := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, _ *http.Request) {
+		w.WriteHeader(http.StatusOK)
+	}))
+	defer backend.Close()
+
+	proxy, err := NewBackendConsoleProxy(backend.URL, nil)
+	require.NoError(t, err)
+	cfg := config.DefaultPublicHTTPConfig()
+	cfg.AntiAbuse.Admin.RateLimit = config.PublicRateLimitConfig{Requests: 1, Window: time.Minute, Burst: 1}
+	handler := newPublicHandlerWithConfig(cfg, ServerDependencies{AdminConsoleProxy: proxy})
+
+	do := func() int {
+		req := proxyRequest(t, http.MethodGet, "http://galaxy.lan/_gm/", nil)
+		req.RemoteAddr = "203.0.113.7:5555"
+		rec := httptest.NewRecorder()
+		handler.ServeHTTP(rec, req)
+		return rec.Code
+	}
+
+	assert.Equal(t, http.StatusOK, do(), "first request within budget")
+	assert.Equal(t, http.StatusTooManyRequests, do(), "second request exhausts the per-IP admin budget")
+}
+
+func TestAdminConsoleProxyReturns502WhenBackendUnreachable(t *testing.T) {
+	backend := httptest.NewServer(http.HandlerFunc(func(http.ResponseWriter, *http.Request) {}))
+	backendURL := backend.URL
+	backend.Close() // close immediately so the next dial is refused
+
+	proxy, err := NewBackendConsoleProxy(backendURL, nil)
+	require.NoError(t, err)
+	handler := newPublicHandlerWithConfig(config.DefaultPublicHTTPConfig(), ServerDependencies{AdminConsoleProxy: proxy})
+
+	req := proxyRequest(t, http.MethodGet, "http://galaxy.lan/_gm/", nil)
+	rec := httptest.NewRecorder()
+	handler.ServeHTTP(rec, req)
+
+	assert.Equal(t, http.StatusBadGateway, rec.Code)
+}
+
+func TestAdminConsoleNotMountedWhenProxyNil(t *testing.T) {
+	handler := newPublicHandler(ServerDependencies{})
+
+	req := proxyRequest(t, http.MethodGet, "http://galaxy.lan/_gm/", nil)
+	rec := httptest.NewRecorder()
+	handler.ServeHTTP(rec, req)
+
+	assert.Equal(t, http.StatusNotFound, rec.Code)
+}
+
+func TestNewBackendConsoleProxyRejectsRelativeURL(t *testing.T) {
+	_, err := NewBackendConsoleProxy("/not-absolute", nil)
+	assert.Error(t, err)
+}

gateway/internal/restapi/public_anti_abuse.go

@@ -234,6 +234,8 @@ func publicRoutePolicyForClass(policy config.PublicHTTPAntiAbuseConfig, class Pu
 		return policy.BrowserBootstrap
 	case PublicRouteClassBrowserAsset:
 		return policy.BrowserAsset
+	case PublicRouteClassAdmin:
+		return policy.Admin
 	default:
 		return policy.PublicMisc
 	}
@@ -252,6 +254,8 @@ func publicAuthIdentityPolicyForPath(requestPath string, policy config.PublicHTT
 
 func allowedMethodsForRequestShape(r *http.Request) []string {
 	switch {
+	case isAdminConsolePath(r.URL.Path):
+		return []string{http.MethodGet, http.MethodHead, http.MethodPost}
 	case isPublicAuthPath(r.URL.Path):
 		return []string{http.MethodPost}
 	case isProbePath(r.URL.Path):
@@ -284,6 +288,17 @@ func isPublicAuthPath(requestPath string) bool {
 	}
 }
 
+// isAdminConsoleRequest reports whether r targets the operator console surface.
+func isAdminConsoleRequest(r *http.Request) bool {
+	return isAdminConsolePath(r.URL.Path)
+}
+
+// isAdminConsolePath reports whether requestPath is the admin console root
+// (`/_gm`) or any path beneath it (`/_gm/...`).
+func isAdminConsolePath(requestPath string) bool {
+	return requestPath == "/_gm" || strings.HasPrefix(requestPath, "/_gm/")
+}
+
 func isProbePath(requestPath string) bool {
 	switch requestPath {
 	case "/healthz", "/readyz":

gateway/internal/restapi/server.go

@@ -48,6 +48,10 @@ const (
 	// PublicRouteClassPublicMisc identifies public traffic that does not match a
 	// more specific class.
 	PublicRouteClassPublicMisc PublicRouteClass = "public_misc"
+
+	// PublicRouteClassAdmin identifies operator console traffic reverse-proxied
+	// to the backend under the `/_gm` prefix.
+	PublicRouteClassAdmin PublicRouteClass = "admin"
 )
 
 var configureGinModeOnce sync.Once
@@ -60,6 +64,7 @@ func (c PublicRouteClass) Normalized() PublicRouteClass {
 	case PublicRouteClassPublicAuth,
 		PublicRouteClassBrowserBootstrap,
 		PublicRouteClassBrowserAsset,
+		PublicRouteClassAdmin,
 		PublicRouteClassPublicMisc:
 		return c
 	default:
@@ -110,6 +115,14 @@ type ServerDependencies struct {
 	// Telemetry records low-cardinality edge metrics. When nil, metrics are
 	// disabled.
 	Telemetry *telemetry.Runtime
+
+	// AdminConsoleProxy, when non-nil, handles `/_gm` and `/_gm/*` by
+	// reverse-proxying to the backend operator console after the public
+	// anti-abuse layer (per-IP rate limit, body, and method checks for the
+	// admin route class) has run. Authentication is delegated to the
+	// backend's admin Basic Auth, whose 401 challenge passes straight back
+	// to the browser. When nil, the admin console surface is not mounted.
+	AdminConsoleProxy http.Handler
 }
 
 // Server owns the public unauthenticated REST listener exposed by the gateway.
@@ -229,6 +242,8 @@ type defaultPublicTrafficClassifier struct{}
 // later drive anti-abuse policy and rate limiting.
 func (defaultPublicTrafficClassifier) Classify(r *http.Request) PublicRouteClass {
 	switch {
+	case isAdminConsoleRequest(r):
+		return PublicRouteClassAdmin
 	case isPublicAuthRequest(r):
 		return PublicRouteClassPublicAuth
 	case isBrowserBootstrapRequest(r):
@@ -290,6 +305,12 @@ func newPublicHandlerWithConfig(cfg config.PublicHTTPConfig, deps ServerDependen
 	router.POST("/api/v1/public/auth/send-email-code", handleSendEmailCode(deps.AuthService, cfg.AuthUpstreamTimeout))
 	router.POST("/api/v1/public/auth/confirm-email-code", handleConfirmEmailCode(deps.AuthService, cfg.AuthUpstreamTimeout))
 
+	if deps.AdminConsoleProxy != nil {
+		adminConsole := gin.WrapH(deps.AdminConsoleProxy)
+		router.Any("/_gm", adminConsole)
+		router.Any("/_gm/*proxyPath", adminConsole)
+	}
+
 	router.NoMethod(func(c *gin.Context) {
 		allowMethods := allowedMethodsForPath(c.Request.URL.Path)
 		if allowMethods != "" {

gateway/internal/restapi/server_test.go

@@ -169,6 +169,31 @@ func TestDefaultPublicTrafficClassifier(t *testing.T) {
 			accept:    "text/html",
 			wantClass: PublicRouteClassBrowserBootstrap,
 		},
+		{
+			name:      "admin console root",
+			method:    http.MethodGet,
+			target:    "/_gm",
+			wantClass: PublicRouteClassAdmin,
+		},
+		{
+			name:      "admin console page wins over browser accept header",
+			method:    http.MethodGet,
+			target:    "/_gm/users",
+			accept:    "text/html",
+			wantClass: PublicRouteClassAdmin,
+		},
+		{
+			name:      "admin console asset wins over browser asset shape",
+			method:    http.MethodGet,
+			target:    "/_gm/assets/console.css",
+			wantClass: PublicRouteClassAdmin,
+		},
+		{
+			name:      "admin console form post",
+			method:    http.MethodPost,
+			target:    "/_gm/users/123/sanctions",
+			wantClass: PublicRouteClassAdmin,
+		},
 	}
 
 	for _, tt := range tests {
@@ -215,6 +240,11 @@ func TestPublicRouteClassNormalized(t *testing.T) {
 			input: PublicRouteClassPublicMisc,
 			want:  PublicRouteClassPublicMisc,
 		},
+		{
+			name:  "admin",
+			input: PublicRouteClassAdmin,
+			want:  PublicRouteClassAdmin,
+		},
 		{
 			name:  "unknown collapses to misc",
 			input: PublicRouteClass("unexpected"),

integration/admin_console_test.go

@@ -0,0 +1,63 @@
+package integration_test
+
+import (
+	"context"
+	"io"
+	"net/http"
+	"strings"
+	"testing"
+	"time"
+
+	"galaxy/integration/testenv"
+)
+
+// TestAdminConsole_ThroughGateway verifies the full operator-console pipe:
+// the edge gateway reverse-proxies `/_gm/*` to the backend, the backend
+// enforces the admin Basic Auth and relays its 401 challenge unchanged, and an
+// authenticated request renders the server-side dashboard.
+func TestAdminConsole_ThroughGateway(t *testing.T) {
+	plat := testenv.Bootstrap(t, testenv.BootstrapOptions{})
+	ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
+	defer cancel()
+
+	base := plat.Gateway.HTTPURL
+	client := &http.Client{Timeout: 10 * time.Second}
+
+	// Unauthenticated: the backend's 401 + Basic challenge must reach the client
+	// through the gateway.
+	req, err := http.NewRequestWithContext(ctx, http.MethodGet, base+"/_gm/", nil)
+	if err != nil {
+		t.Fatalf("build request: %v", err)
+	}
+	resp, err := client.Do(req)
+	if err != nil {
+		t.Fatalf("GET /_gm/ (no auth): %v", err)
+	}
+	body, _ := io.ReadAll(resp.Body)
+	resp.Body.Close()
+	if resp.StatusCode != http.StatusUnauthorized {
+		t.Fatalf("no-auth status = %d, want 401; body=%s", resp.StatusCode, body)
+	}
+	if !strings.Contains(resp.Header.Get("WWW-Authenticate"), "Basic") {
+		t.Fatalf("WWW-Authenticate = %q, want a Basic challenge", resp.Header.Get("WWW-Authenticate"))
+	}
+
+	// Authenticated with the bootstrap operator: the dashboard renders.
+	req, err = http.NewRequestWithContext(ctx, http.MethodGet, base+"/_gm/", nil)
+	if err != nil {
+		t.Fatalf("build request: %v", err)
+	}
+	req.SetBasicAuth(plat.Backend.AdminUser, plat.Backend.AdminPassword)
+	resp, err = client.Do(req)
+	if err != nil {
+		t.Fatalf("GET /_gm/ (auth): %v", err)
+	}
+	body, _ = io.ReadAll(resp.Body)
+	resp.Body.Close()
+	if resp.StatusCode != http.StatusOK {
+		t.Fatalf("authenticated status = %d, want 200; body=%s", resp.StatusCode, body)
+	}
+	if !strings.Contains(string(body), "Dashboard") {
+		t.Fatalf("dashboard body missing the heading; got: %s", body)
+	}
+}

tools/dev-deploy/.env.example

@@ -7,12 +7,6 @@
 # baked into `docker-compose.yml`, so this file documents the knobs
 # rather than driving them.
 
-# Auto-provisioned sandbox bootstrap. Empty disables the bootstrap.
-BACKEND_DEV_SANDBOX_EMAIL=dev@galaxy.lan
-BACKEND_DEV_SANDBOX_ENGINE_IMAGE=galaxy-engine:dev
-BACKEND_DEV_SANDBOX_ENGINE_VERSION=0.1.0
-BACKEND_DEV_SANDBOX_PLAYER_COUNT=20
-
 # `123456` short-circuits the email-code path for the dev account.
 # This is also the docker-compose default — set the variable to an
 # empty string here when the environment must rely on real Mailpit

tools/dev-deploy/Caddyfile.dev

@@ -29,6 +29,36 @@
 		reverse_proxy galaxy-api:8080
 	}
 
+	# Operator console + observability behind one Basic Auth gate. The gate
+	# credential equals the admin-console account (dev: gm / gm-dev-password),
+	# so Caddy forwards the same Authorization header to the backend `/_gm`
+	# surface (its own Basic Auth) and to Grafana/Mailpit — one prompt covers
+	# all three. The gateway applies the admin anti-abuse class to the console.
+	@gm path /_gm /_gm/*
+	handle @gm {
+		basic_auth {
+			gm "$2a$14$xVh1TLaZxh8fazlKrI9Mx.NQMQlMarYWtr3FRELmZIXuac/DeeTRO"
+		}
+
+		# Grafana under /_gm/grafana/ (sub-path mode; anonymous Admin, so the
+		# /_gm gate is the only barrier — GF_AUTH_BASIC_ENABLED=false makes it
+		# ignore the forwarded Authorization header).
+		handle /_gm/grafana/* {
+			reverse_proxy galaxy-grafana:3000
+		}
+
+		# Mailpit captured-mail UI under /_gm/mailpit/ (MP_WEBROOT). Shows
+		# every message the backend sent, relayed or not.
+		handle /_gm/mailpit/* {
+			reverse_proxy galaxy-mailpit:8025
+		}
+
+		# The operator console itself (gateway -> backend /_gm surface).
+		handle {
+			reverse_proxy galaxy-api:8080
+		}
+	}
+
 	# Bare `/game` (no trailing slash) -> `/game/` so the SPA root
 	# resolves before the site catch-all can claim it.
 	handle /game {

tools/dev-deploy/KNOWN-ISSUES.md

@@ -1,164 +1,8 @@
 # `tools/dev-deploy/` — known issues
 
-Issues that surface in the long-lived dev environment but are not yet
+Issues that surfaced in the long-lived dev environment. Each entry lists
-fixed. Each entry lists the observed symptom, the diagnostic evidence,
+the observed symptom, the diagnostic evidence, and the fix or the open
-the working hypothesis, and the open questions that have to be
+questions that have to be answered before a fix lands.
-answered before a fix lands.
-
-## Dev Sandbox game flips to `cancelled` after a `dev-deploy` redispatch
-
-### Symptom
-
-A previously `running` "Dev Sandbox" game (created by
-`backend/internal/devsandbox`) transitions to `cancelled` ~15 minutes
-after a `dev-deploy.yaml` workflow_dispatch run finishes. The user's
-browser session survives (the same `device_session_id` keeps working),
-but the lobby shows no game because the only game it had is now
-terminal. `purgeTerminalSandboxGames` does pick it up on the **next**
-boot and creates a fresh sandbox — but the first redispatch leaves
-the user with an empty lobby until backend restarts again.
-
-### Diagnostic evidence
-
-Backend logs from the broken cycle (timestamps abbreviated):
-
-```text
-20:24:40 dev_sandbox: purged terminal sandbox game game_id=<prev> status=cancelled
-20:24:40 dev_sandbox: memberships ensured count=20 game_id=<new>
-20:24:40 dev_sandbox: bootstrap complete user_id=<owner> game_id=<new> status=starting
-...
-20:25:09 user mail sent failed (diplomail tables missing — unrelated)
-...
-20:39:40 lobby: game cancelled by runtime reconciler game_id=<new>
-         op=reconcile status=removed message="container disappeared"
-```
-
-Between 20:24:40 (`status=starting`) and 20:39:40 (reconciler cancel)
-the backend logs are silent on the runtime / engine paths — no
-`engine spawned`, no `engine container started`, no `runtime
-transition` lines. The reconciler then fires and reports the engine
-container as missing.
-
-`docker ps -a --filter 'label=org.opencontainers.image.title=galaxy-game-engine'`
-returns no rows during this window — the engine container is neither
-running nor stopped on the host, so it either was never spawned or
-was removed before the host snapshot.
-
-### What has been ruled out
-
-A live `docker inspect` on a healthy engine container shows:
-
-```text
-Labels: galaxy.backend=1, galaxy.engine_version=0.1.0,
-        galaxy.game_id=<uuid>,
-        org.opencontainers.image.title=galaxy-game-engine,
-        com.galaxy.{cpu_quota,memory,pids_limit}
-AutoRemove:    false
-RestartPolicy: on-failure
-NetworkMode:   galaxy-dev-internal
-```
-
-There are no `com.docker.compose.*` labels and `AutoRemove=false`,
-so `--remove-orphans` cannot reap the engine and a `--rm`-style
-self-destruct is not in play. Two redispatches captured under
-`docker events --filter event=create,start,die,destroy,kill,stop`
-also confirmed it: across both runs the only `die` / `destroy`
-events were for `galaxy-dev-{backend,api,caddy}`. The live engine
-container survived both redispatches, and the reconciler that
-fires 60 seconds after the new backend boots correctly matched
-it through `byGameID` / `byContainerID`.
-
-`backend/internal/runtime/service.go` only removes engine
-containers from the explicit `runStop` / `runRestart` / `runPatch`
-paths. There is no `runtime.Service.Shutdown` that proactively
-kills containers on backend exit, so a graceful SIGTERM to
-`galaxy-dev-backend` will not touch its child engine containers.
-
-### Host-side hypotheses considered and rejected by the owner
-
-The natural follow-up suspects after compose was cleared — host-side
-`docker prune` cron jobs, a manual `docker rm`, an out-of-band
-`dockerd` restart, and an idle-state engine crash — were all
-rejected by the project owner: the dev host runs none of those
-periodic cleanups, no one manually removed the container, dockerd
-was not restarted in the window, and the engine binary does not
-crash while idling on API calls.
-
-### Best remaining suspicion
-
-Something the `dev-deploy.yaml` CI run does between successful
-image builds and the final `docker compose up -d --wait
---remove-orphans` clobbers the previously-spawned engine container.
-The chain at runtime contains:
-
-1. `docker build -t galaxy-engine:dev -f game/Dockerfile .`
-2. `docker compose build galaxy-backend galaxy-api`
-3. `docker run --rm` alpine for the UI volume seed
-4. `docker compose up -d --wait --remove-orphans`
-
-None of these *should* touch an unmanaged engine container, but
-the reproduction window points squarely inside this sequence. A
-deliberate next reproduction with `docker events --since 0` armed
-*before* the deploy starts and live for the entire job — captured
-end-to-end on the dev host, not just the chunk after backend
-recreate — would pin which step emits the `destroy` on the engine.
-
-### Update 2026-05-19: integration preclean identified as one cause
-
-A live reproduction during the post-merge auto-deploy cycle (Gitea
-run #188 dev-deploy plus parallel run #190 integration) pinned one
-clobbering source: `integration/scripts/preclean.sh` was unscoped
-and removed *every* container labelled `galaxy.backend=1`, including
-the dev-deploy engine. Timeline from the dev host:
-
-```text
-23:10:40  backend pre-bootstrap reconciler tick: engine alive
-23:10:40  dev_sandbox bootstrap: status=running
-23:10:56  preclean: removing 1 backend-managed engine containers  ← integration run #190
-23:11:40  reconciler: container disappeared → game cancelled
-```
-
-Fix landed: `BACKEND_STACK_LABEL=integration` is now passed to
-every integration backend (see
-`integration/testenv/backend.go`) and `preclean.sh` AND-combines
-`galaxy.backend=1` with `galaxy.stack=integration`, so dev-deploy /
-local-dev engines stamped with different stack values are no longer
-collateral.
-
-This covers **push**-triggered cycles where `dev-deploy.yaml` and
-`integration.yaml` run on the same Gitea host. The original
-hypothesis (a `workflow_dispatch dev-deploy` solo run also losing
-the engine) is *not* explained by the integration fix — manual
-dispatches do not trigger `integration.yaml`. Keep this entry open
-until a solo-dispatch reproduction confirms whether the symptom
-still occurs.
-
-### Status
-
-Partially fixed (push-triggered cycles). Solo `workflow_dispatch`
-reproductions still open. If the symptom recurs after the
-integration fix lands, capture `docker events --since 0` for the
-full dispatch window and attach here.
-
-### Workaround in use today
-
-When the sandbox game flips to `cancelled`, redispatch `dev-deploy`:
-
-```sh
-curl -X POST -n -H 'Content-Type: application/json' \
-  -d '{"ref":"<branch>"}' \
-  https://gitea.iliadenisov.ru/api/v1/repos/developer/galaxy-game/actions/workflows/dev-deploy.yaml/dispatches
-```
-
-The next boot's `purgeTerminalSandboxGames` removes the cancelled
-row, `findOrCreateSandboxGame` creates a fresh one, and
-`ensureMembershipsAndDrive` puts the new game back to `running`.
-
-### Owner
-
-Unassigned. File an issue once we have the runtime / reconciler
-analysis above; reference this section in the issue body so future
-redeploys can short-circuit the diagnostic loop.
 
 ## `docker restart galaxy-dev-backend` fails after the CI runner cleans up
 

tools/dev-deploy/README.md

@@ -114,17 +114,72 @@ calls `make clean-data`.
 The same dev-mode email-code override as `tools/local-dev/` applies,
 and the dev-deploy compose ships with it enabled by default:
 
-1. Enter `dev@galaxy.lan` (or whatever `BACKEND_DEV_SANDBOX_EMAIL`
+1. Enter your email address in the login form.
-   resolves to) in the login form.
 2. Submit `123456` as the code — the docker-compose default for
    `BACKEND_AUTH_DEV_FIXED_CODE` is `123456`, so the bcrypt-hashed
-   email code stays a fallback. To force real Mailpit codes (e.g. for
+   email code stays a fallback. To force the real email code (which
-   mail-flow QA), set `BACKEND_AUTH_DEV_FIXED_CODE=` (empty) in a
+   Mailpit then relays to your Gmail — see **Mail** below), set
-   local `.env` and `make rebuild`.
+   `BACKEND_AUTH_DEV_FIXED_CODE=` (empty) and redeploy.
 
 The fixed-code override is rejected by production env loaders, so it
 cannot leak into the prod environment.
 
+## Mail
+
+The backend always submits mail to **Mailpit** (`galaxy-mailpit:1025`),
+exactly as it would to a production SMTP server. Mailpit captures every
+message in its UI (internal `:8025`) and, when configured, **relays**
+the ones whose recipient matches `GALAXY_DEV_MAIL_RELAY_MATCH` up to a
+real Gmail account — so an OTP addressed to you lands in your real inbox
+while everything else stays captured-only.
+
+Configure the relay through Gitea Actions secrets/vars (never
+committed); the `dev-deploy.yaml` workflow renders Mailpit's
+`relay.conf` (from `tools/dev-deploy/mailpit/relay.conf.tmpl`) and seeds
+it into the `galaxy-dev-mailpit-config` volume:
+
+| Name | Kind | Purpose |
+| --- | --- | --- |
+| `GALAXY_DEV_MAIL_RELAY_USERNAME` | secret | Gmail address used as the relay login + From. |
+| `GALAXY_DEV_MAIL_RELAY_PASSWORD` | secret | Gmail **App Password** (requires 2FA; not the account password). |
+| `GALAXY_DEV_MAIL_RELAY_MATCH` | var | Recipient regex to auto-relay (e.g. your Gmail address). Unset → capture-only. |
+
+With none set the stack only captures mail (the compose relay-match
+defaults to a non-routable address), so it can never email third
+parties.
+
+The capture UI is exposed through the operator console's `/_gm` gate at
+[`/_gm/mailpit/`](https://galaxy.lan/_gm/mailpit/) — one Basic Auth for
+the console, Grafana and Mailpit (see **Observability**). It shows
+**every** message the backend sent, relayed or not, so you can read any
+account's OTP regardless of the relay-match. For multi-account testing:
+register several `you+tag@gmail.com` aliases and widen the match to a
+regex such as `^you(\+[^@]+)?@gmail\.com$` (Gmail folds every `+tag`
+into one inbox), or just read the codes in the Mailpit UI, or skip mail
+entirely with the `123456` dev-code.
+
+## Observability
+
+A full metrics + logs + traces stack runs alongside the app on the
+internal network (no host ports), as a production mirror. **Grafana**
+and the **Mailpit** UI are reached only through the operator console's
+single `/_gm` Basic Auth gate — one password (the admin-console account)
+unlocks the console, [`/_gm/grafana/`](https://galaxy.lan/_gm/grafana/)
+and [`/_gm/mailpit/`](https://galaxy.lan/_gm/mailpit/), with links in the
+console nav. Grafana runs anonymous-Admin behind the gate (no own
+login); Prometheus, Loki and Tempo stay internal-only.
+
+- **Metrics** — Prometheus scrapes backend, gateway, `node-exporter` and
+  cAdvisor.
+- **Logs** — promtail → Loki (Docker SD on the `galaxy.stack=dev-deploy`
+  label).
+- **Traces** — backend + gateway → Tempo over OTLP.
+
+Grafana's admin user is seeded from `GALAXY_DEV_GRAFANA_ADMIN_PASSWORD`
+(for provisioning/API; the UI needs no Grafana login). See
+[`monitoring/README.md`](monitoring/README.md) for services, configs and
+tuning knobs.
+
 ## Networking
 
 ```
@@ -139,6 +194,8 @@ galaxy-caddy  (networks: edge + galaxy-dev-internal)
    │  /game/*          -> file_server /srv/galaxy-ui   (volume galaxy-dev-ui-dist)
    │  /api/*, /healthz -> reverse_proxy galaxy-api:8080
    │  /rpc/*           -> reverse_proxy galaxy-api:9090 (strips /rpc)
+   │  /_gm, /_gm/*     -> reverse_proxy galaxy-api:8080 (Basic Auth gate;
+   │                      /_gm/grafana/ -> grafana, /_gm/mailpit/ -> mailpit)
    ▼
 galaxy-dev-internal
    ├─ galaxy-api      (gateway:   :8080 REST, :9090 gRPC)
@@ -146,7 +203,9 @@ galaxy-dev-internal
    ├─ galaxy-postgres (postgres:  :5432)
    ├─ galaxy-redis    (redis:     :6379)
    ├─ galaxy-mailpit  (mailpit:   :8025 UI, :1025 SMTP)
-   └─ engine containers (spawned by backend on demand)
+   ├─ engine containers (spawned by backend on demand)
+   └─ observability   (prometheus, grafana, loki, promtail, tempo,
+                       node-exporter, cadvisor)
 ```
 
 The compose project deliberately exposes no host ports. Diagnostics
@@ -191,8 +250,10 @@ make clean-data     Stop everything and wipe volumes + game-state dir
 
 ## Files
 
-- `docker-compose.yml` — six services: postgres, redis, mailpit,
+- `docker-compose.yml` — the application services (postgres, redis,
-  galaxy-backend, galaxy-api, galaxy-caddy. `galaxy-caddy` mounts both
+  mailpit, galaxy-backend, galaxy-api, galaxy-caddy) plus the
+  observability stack (prometheus, grafana, loki, promtail, tempo,
+  node-exporter, cadvisor). `galaxy-caddy` mounts both
   the `galaxy-dev-site-dist` (`/srv/galaxy-site`) and
   `galaxy-dev-ui-dist` (`/srv/galaxy-ui`) volumes and reverse-proxies
   both gateway tiers (REST/health on `:8080`, Connect/gRPC-web on
@@ -204,6 +265,8 @@ make clean-data     Stop everything and wipe volumes + game-state dir
   at `/etc/caddy/Caddyfile`.
 - `Caddyfile.prod` — placeholder for a future prod deployment; not used
   by this compose.
+- `monitoring/` — Prometheus / Loki / promtail / Tempo / Grafana
+  configuration, provisioned as code; see `monitoring/README.md`.
 - `Makefile` — wrapper over `docker compose` with helpers for engine,
   site/UI seeding, health probes, and full wipe.
 - `.env.example` — non-secret defaults for the compose `${VAR:-}`
@@ -212,8 +275,7 @@ make clean-data     Stop everything and wipe volumes + game-state dir
 ## Known issues
 
 See [`KNOWN-ISSUES.md`](KNOWN-ISSUES.md) for symptoms that surface
-in the long-lived dev environment but are not yet fixed (currently:
+in the long-lived dev environment but are not yet fixed.
-the sandbox game flipping to `cancelled` after a redispatch).
 
 ## Deployment cadence
 
@@ -237,12 +299,12 @@ behind. There is no separate state to clean up between the two paths.
 
 ### Engine image drift recycle
 
-`backend` spawns one engine container per game (the long-lived "Dev
+`backend` spawns one engine container per running game and the
-Sandbox" plus any user-created games) and the reconciler reattaches
+reconciler reattaches to whatever it finds with the
-to whatever it finds with the `galaxy.stack=dev-deploy` label. That
+`galaxy.stack=dev-deploy` label. That reattach does not check the
-reattach does not check the running container's image SHA against the
+running container's image SHA against the freshly-built
-freshly-built `galaxy-engine:dev` tag, so an unchanged container would
+`galaxy-engine:dev` tag, so an unchanged container would otherwise
-otherwise keep serving the previous engine code after a redeploy.
+keep serving the previous engine code after a redeploy.
 
 The `dev-deploy.yaml` workflow handles this in the
 `Recycle engine containers on image drift` step. When `docker build`
@@ -250,9 +312,7 @@ produces a new `galaxy-engine:dev` SHA, the step compares it against
 every running `galaxy-game-*` container and, for each drifted one,
 stops the backend, removes the container, wipes its bind-mounted
 state directory (Engine.Init() writes turn-0 over any pre-existing
-`turn-N` files), and cascade-deletes the lobby `games` row. The
+`turn-N` files), and cascade-deletes the lobby `games` row.
-`dev-sandbox` bootstrap on the next backend boot finds no live
-sandbox and provisions a fresh one on the new engine image.
 
 When the engine sources are unchanged, the BuildKit cache hits and
 the SHA stays the same — the recycle step is a no-op and the running

tools/dev-deploy/docker-compose.yml

@@ -66,12 +66,26 @@ services:
     image: axllent/mailpit:v1.21
     container_name: galaxy-dev-mailpit
     restart: unless-stopped
+    # Mailpit is both the SMTP submission point and a relay: it captures
+    # every message in its UI and auto-relays the ones whose recipient
+    # matches GALAXY_DEV_MAIL_RELAY_MATCH to the Gmail account in the
+    # secret-rendered relay config. The default match is non-routable, so
+    # a stack brought up without the relay secret only captures, never sends.
+    command:
+      - "--smtp-relay-config=/etc/mailpit/relay.conf"
+      - "--smtp-relay-matching=${GALAXY_DEV_MAIL_RELAY_MATCH:-nobody@invalid.example}"
+      # Serve the capture UI under /_gm/mailpit so the host Caddy can expose
+      # it at https://galaxy.lan/_gm/mailpit/ behind the shared /_gm gate;
+      # SMTP is unaffected.
+      - "--webroot=/_gm/mailpit"
     labels:
       galaxy.stack: dev-deploy
     networks:
       - galaxy-internal
+    volumes:
+      - galaxy-dev-mailpit-config:/etc/mailpit:ro
     healthcheck:
-      test: ["CMD", "wget", "-q", "-O-", "http://localhost:8025/livez"]
+      test: ["CMD", "wget", "-q", "-O-", "http://localhost:8025/_gm/mailpit/livez"]
       interval: 3s
       timeout: 3s
       retries: 30
@@ -108,23 +122,31 @@ services:
       BACKEND_NOTIFICATION_ADMIN_EMAIL: admin@galaxy.lan
       BACKEND_MAIL_WORKER_INTERVAL: 500ms
       BACKEND_NOTIFICATION_WORKER_INTERVAL: 500ms
-      BACKEND_OTEL_TRACES_EXPORTER: none
+      BACKEND_OTEL_TRACES_EXPORTER: otlp
-      BACKEND_OTEL_METRICS_EXPORTER: none
+      BACKEND_OTEL_PROTOCOL: grpc
+      BACKEND_OTEL_ENDPOINT: "galaxy-tempo:4317"
+      # Tempo's OTLP receiver is plaintext on the internal network; the
+      # backend's gRPC exporter defaults to TLS, so disable it via the
+      # standard SDK env (applied on top of WithEndpoint).
+      OTEL_EXPORTER_OTLP_INSECURE: "true"
+      # Prometheus metrics are enabled in dev so the `/metrics` scrape
+      # endpoint is live and stable ahead of standing up a Prometheus +
+      # Grafana stack on the internal network. The listener stays internal
+      # (not mapped to the host); nothing scrapes it yet.
+      BACKEND_OTEL_METRICS_EXPORTER: prometheus
+      BACKEND_OTEL_PROMETHEUS_LISTEN_ADDR: ":9100"
+      # Operator console (`/_gm`): Basic Auth bootstrap account plus the
+      # stateless CSRF key. Dev-only non-secrets, overridable via `.env`; a
+      # stable CSRF key keeps console forms valid across redeploys.
+      BACKEND_ADMIN_BOOTSTRAP_USER: ${BACKEND_ADMIN_BOOTSTRAP_USER:-gm}
+      BACKEND_ADMIN_BOOTSTRAP_PASSWORD: ${BACKEND_ADMIN_BOOTSTRAP_PASSWORD:-gm-dev-password}
+      BACKEND_ADMIN_CONSOLE_CSRF_KEY: ${BACKEND_ADMIN_CONSOLE_CSRF_KEY:-dev-admin-console-csrf-key}
       # Long-lived dev environment always opts into the fixed-code
       # override so a returning developer can sign in with `123456`
       # even after the matching browser session was cleared (the real
       # bcrypt-hashed code is single-use). Set the var to an empty
       # string in `.env` to disable.
       BACKEND_AUTH_DEV_FIXED_CODE: ${BACKEND_AUTH_DEV_FIXED_CODE:-123456}
-      # Long-lived dev environment always bootstraps the "Dev Sandbox"
-      # game owned by this email so a freshly redeployed stack already
-      # has one ready-to-play game in the lobby. Set the variable to an
-      # empty string in `.env` to disable the bootstrap (e.g. for a
-      # cold-start QA pass).
-      BACKEND_DEV_SANDBOX_EMAIL: ${BACKEND_DEV_SANDBOX_EMAIL:-dev@galaxy.lan}
-      BACKEND_DEV_SANDBOX_ENGINE_IMAGE: ${BACKEND_DEV_SANDBOX_ENGINE_IMAGE:-galaxy-engine:dev}
-      BACKEND_DEV_SANDBOX_ENGINE_VERSION: ${BACKEND_DEV_SANDBOX_ENGINE_VERSION:-0.1.0}
-      BACKEND_DEV_SANDBOX_PLAYER_COUNT: ${BACKEND_DEV_SANDBOX_PLAYER_COUNT:-20}
     volumes:
       - /var/run/docker.sock:/var/run/docker.sock
       # Per-game state directories live under the same absolute path
@@ -180,6 +202,16 @@ services:
       GATEWAY_LOG_LEVEL: info
       GATEWAY_PUBLIC_HTTP_ADDR: ":8080"
       GATEWAY_AUTHENTICATED_GRPC_ADDR: ":9090"
+      # Private admin listener exposes the Prometheus `/metrics` endpoint on
+      # the internal network — live and stable for a future scrape, not
+      # mapped to the host.
+      GATEWAY_ADMIN_HTTP_ADDR: ":9191"
+      # Traces -> Tempo over OTLP gRPC (plaintext on the internal net).
+      OTEL_SERVICE_NAME: galaxy-gateway
+      OTEL_TRACES_EXPORTER: otlp
+      OTEL_EXPORTER_OTLP_PROTOCOL: grpc
+      OTEL_EXPORTER_OTLP_ENDPOINT: "http://galaxy-tempo:4317"
+      OTEL_EXPORTER_OTLP_INSECURE: "true"
       GATEWAY_BACKEND_HTTP_URL: "http://galaxy-backend:8080"
       GATEWAY_BACKEND_GRPC_PUSH_URL: "galaxy-backend:8081"
       GATEWAY_BACKEND_GATEWAY_CLIENT_ID: dev-gateway-1
@@ -208,6 +240,9 @@ services:
       GATEWAY_PUBLIC_HTTP_ANTI_ABUSE_PUBLIC_MISC_RATE_LIMIT_BURST: "1000"
       GATEWAY_PUBLIC_HTTP_ANTI_ABUSE_BROWSER_BOOTSTRAP_MAX_BODY_BYTES: "65536"
       GATEWAY_PUBLIC_HTTP_ANTI_ABUSE_BROWSER_ASSET_MAX_BODY_BYTES: "65536"
+      GATEWAY_PUBLIC_HTTP_ANTI_ABUSE_ADMIN_MAX_BODY_BYTES: "131072"
+      GATEWAY_PUBLIC_HTTP_ANTI_ABUSE_ADMIN_RATE_LIMIT_REQUESTS: "10000"
+      GATEWAY_PUBLIC_HTTP_ANTI_ABUSE_ADMIN_RATE_LIMIT_BURST: "1000"
       GATEWAY_AUTHENTICATED_GRPC_ANTI_ABUSE_IP_RATE_LIMIT_REQUESTS: "10000"
       GATEWAY_AUTHENTICATED_GRPC_ANTI_ABUSE_IP_RATE_LIMIT_BURST: "1000"
       GATEWAY_AUTHENTICATED_GRPC_ANTI_ABUSE_SESSION_RATE_LIMIT_REQUESTS: "10000"
@@ -245,6 +280,163 @@ services:
       - galaxy-internal
       - edge
 
+  galaxy-prometheus:
+    image: prom/prometheus:v2.55.1
+    container_name: galaxy-dev-prometheus
+    restart: unless-stopped
+    labels:
+      galaxy.stack: dev-deploy
+    command:
+      - --config.file=/etc/prometheus/prometheus.yml
+      - --storage.tsdb.path=/prometheus
+      - --storage.tsdb.retention.time=15d
+      - --web.enable-lifecycle
+    volumes:
+      - ${GALAXY_DEV_MONITORING_DIR:-./monitoring}/prometheus/prometheus.yml:/etc/prometheus/prometheus.yml:ro
+      - galaxy-dev-prometheus-data:/prometheus
+    networks:
+      - galaxy-internal
+    deploy:
+      resources:
+        limits:
+          memory: 384m
+
+  galaxy-loki:
+    image: grafana/loki:3.3.2
+    container_name: galaxy-dev-loki
+    restart: unless-stopped
+    labels:
+      galaxy.stack: dev-deploy
+    command: ["-config.file=/etc/loki/loki.yml"]
+    volumes:
+      - ${GALAXY_DEV_MONITORING_DIR:-./monitoring}/loki/loki.yml:/etc/loki/loki.yml:ro
+      - galaxy-dev-loki-data:/loki
+    networks:
+      - galaxy-internal
+    deploy:
+      resources:
+        limits:
+          memory: 384m
+
+  galaxy-promtail:
+    image: grafana/promtail:3.3.2
+    container_name: galaxy-dev-promtail
+    restart: unless-stopped
+    labels:
+      galaxy.stack: dev-deploy
+    command: ["-config.file=/etc/promtail/promtail.yml"]
+    volumes:
+      - ${GALAXY_DEV_MONITORING_DIR:-./monitoring}/promtail/promtail.yml:/etc/promtail/promtail.yml:ro
+      - /var/run/docker.sock:/var/run/docker.sock:ro
+    networks:
+      - galaxy-internal
+    deploy:
+      resources:
+        limits:
+          memory: 128m
+
+  galaxy-tempo:
+    image: grafana/tempo:2.7.1
+    container_name: galaxy-dev-tempo
+    restart: unless-stopped
+    labels:
+      galaxy.stack: dev-deploy
+    command: ["-config.file=/etc/tempo/tempo.yml"]
+    volumes:
+      - ${GALAXY_DEV_MONITORING_DIR:-./monitoring}/tempo/tempo.yml:/etc/tempo/tempo.yml:ro
+      - galaxy-dev-tempo-data:/var/tempo
+    networks:
+      - galaxy-internal
+    deploy:
+      resources:
+        limits:
+          memory: 384m
+
+  galaxy-node-exporter:
+    image: prom/node-exporter:v1.8.2
+    container_name: galaxy-dev-node-exporter
+    restart: unless-stopped
+    labels:
+      galaxy.stack: dev-deploy
+    command:
+      - --path.procfs=/host/proc
+      - --path.sysfs=/host/sys
+      - --path.rootfs=/rootfs
+      - --collector.filesystem.mount-points-exclude=^/(sys|proc|dev|host|etc)($$|/)
+    volumes:
+      - /proc:/host/proc:ro
+      - /sys:/host/sys:ro
+      - /:/rootfs:ro
+    pid: host
+    networks:
+      - galaxy-internal
+    deploy:
+      resources:
+        limits:
+          memory: 64m
+
+  galaxy-cadvisor:
+    image: gcr.io/cadvisor/cadvisor:v0.49.1
+    container_name: galaxy-dev-cadvisor
+    restart: unless-stopped
+    labels:
+      galaxy.stack: dev-deploy
+    command:
+      - --housekeeping_interval=30s
+      - --docker_only=true
+      - --store_container_labels=false
+    privileged: true
+    devices:
+      - /dev/kmsg
+    volumes:
+      - /:/rootfs:ro
+      - /var/run:/var/run:ro
+      - /sys:/sys:ro
+      - /var/lib/docker/:/var/lib/docker:ro
+      - /dev/disk/:/dev/disk:ro
+    networks:
+      - galaxy-internal
+    deploy:
+      resources:
+        limits:
+          memory: 256m
+
+  galaxy-grafana:
+    image: grafana/grafana:11.4.0
+    container_name: galaxy-dev-grafana
+    restart: unless-stopped
+    labels:
+      galaxy.stack: dev-deploy
+    depends_on:
+      - galaxy-prometheus
+      - galaxy-loki
+      - galaxy-tempo
+    environment:
+      GF_SECURITY_ADMIN_PASSWORD: ${GALAXY_DEV_GRAFANA_ADMIN_PASSWORD:-admin}
+      GF_SERVER_ROOT_URL: https://galaxy.lan/_gm/grafana/
+      GF_SERVER_SERVE_FROM_SUB_PATH: "true"
+      # No own login: the /_gm Basic Auth gate is the only barrier, so
+      # serve everyone as anonymous Admin and ignore the forwarded
+      # Authorization header (basic auth off, login form off).
+      GF_AUTH_ANONYMOUS_ENABLED: "true"
+      GF_AUTH_ANONYMOUS_ORG_ROLE: Admin
+      GF_AUTH_DISABLE_LOGIN_FORM: "true"
+      GF_AUTH_BASIC_ENABLED: "false"
+      GF_USERS_ALLOW_SIGN_UP: "false"
+      GF_ANALYTICS_REPORTING_ENABLED: "false"
+      GF_ANALYTICS_CHECK_FOR_UPDATES: "false"
+      GF_NEWS_NEWS_FEED_ENABLED: "false"
+    volumes:
+      - ${GALAXY_DEV_MONITORING_DIR:-./monitoring}/grafana/provisioning:/etc/grafana/provisioning:ro
+      - ${GALAXY_DEV_MONITORING_DIR:-./monitoring}/grafana/dashboards:/var/lib/grafana/dashboards:ro
+      - galaxy-dev-grafana-data:/var/lib/grafana
+    networks:
+      - galaxy-internal
+    deploy:
+      resources:
+        limits:
+          memory: 256m
+
 networks:
   galaxy-internal:
     name: galaxy-dev-internal
@@ -274,3 +466,13 @@ volumes:
     name: galaxy-dev-site-dist
   galaxy-dev-geoip-data:
     name: galaxy-dev-geoip-data
+  galaxy-dev-mailpit-config:
+    name: galaxy-dev-mailpit-config
+  galaxy-dev-prometheus-data:
+    name: galaxy-dev-prometheus-data
+  galaxy-dev-grafana-data:
+    name: galaxy-dev-grafana-data
+  galaxy-dev-loki-data:
+    name: galaxy-dev-loki-data
+  galaxy-dev-tempo-data:
+    name: galaxy-dev-tempo-data

tools/dev-deploy/mailpit/relay.conf.tmpl

@@ -0,0 +1,18 @@
+# Mailpit SMTP relay upstream — RENDERED AT DEPLOY TIME by
+# .gitea/workflows/dev-deploy.yaml from Gitea Actions secrets, then
+# seeded into the `galaxy-dev-mailpit-config` volume. The Gmail App
+# Password is a secret and MUST NOT be committed: this template only
+# carries ${PLACEHOLDER}s that the workflow substitutes. See
+# tools/dev-deploy/README.md ("Mail").
+#
+# Mailpit captures every message; the `--smtp-relay-matching` flag (set
+# from GALAXY_DEV_MAIL_RELAY_MATCH in the compose) decides which
+# recipients are actually relayed up to this Gmail account.
+host: smtp.gmail.com
+port: 587
+starttls: true
+allow-insecure: false
+auth: login
+username: ${GALAXY_DEV_MAIL_RELAY_USERNAME}
+password: ${GALAXY_DEV_MAIL_RELAY_PASSWORD}
+return-path: ${GALAXY_DEV_MAIL_RELAY_USERNAME}

tools/dev-deploy/monitoring/README.md

@@ -0,0 +1,77 @@
+# `tools/dev-deploy/monitoring/` — observability stack
+
+The long-lived dev environment runs a full metrics + logs + traces stack
+alongside the application as a **production mirror**: the same compose
+fragment and collector configs are meant to back production later. Every
+collector lives on the internal `galaxy-dev-internal` network and
+publishes **no host port**. The browser-reachable pieces (Grafana and
+the Mailpit UI) sit behind the operator console's single `/_gm` Basic
+Auth gate — see [`../README.md`](../README.md) and `ARCHITECTURE.md §14`.
+
+## Services
+
+| Service | Image | Role | Reachable |
+| --- | --- | --- | --- |
+| `galaxy-prometheus` | `prom/prometheus` | Scrape + store metrics (15d) | internal `:9090` |
+| `galaxy-loki` | `grafana/loki` | Log store (7d) | internal `:3100` |
+| `galaxy-promtail` | `grafana/promtail` | Ship container logs to Loki | — |
+| `galaxy-tempo` | `grafana/tempo` | Trace store (3d), OTLP receiver | internal `:3200`, OTLP `:4317`/`:4318` |
+| `galaxy-node-exporter` | `prom/node-exporter` | Host metrics | internal `:9100` |
+| `galaxy-cadvisor` | `cadvisor` | Per-container CPU/memory/IO | internal `:8080` |
+| `galaxy-grafana` | `grafana/grafana` | Dashboards + Explore | Caddy `/_gm/grafana/` |
+
+## What is collected
+
+- **Metrics.** Prometheus (30s interval) scrapes the backend Prometheus
+  endpoint (`galaxy-backend:9100`), the gateway admin endpoint
+  (`galaxy-api:9191`), `node-exporter` (host) and cAdvisor (per
+  container). Engine containers expose no `/metrics`; cAdvisor covers
+  their resource use.
+- **Logs.** promtail discovers containers through the Docker API,
+  filtered to the `galaxy.stack=dev-deploy` label, and ships their
+  stdout/stderr to Loki labelled by `container`.
+- **Traces.** backend and gateway export OTLP traces over gRPC to Tempo
+  (`galaxy-tempo:4317`), plaintext on the internal network
+  (`OTEL_EXPORTER_OTLP_INSECURE=true`, since Tempo's receiver is not
+  TLS-wrapped inside the contour).
+
+## Grafana access (behind the `/_gm` gate)
+
+Grafana is served under `/_gm/grafana/` (`GF_SERVER_ROOT_URL` +
+`GF_SERVER_SERVE_FROM_SUB_PATH=true`) **behind the shared operator gate**:
+the Caddy `/_gm/*` Basic Auth (the admin-console account) is the only
+barrier. Grafana itself runs as **anonymous Admin** with its login form
+and basic auth disabled (`GF_AUTH_ANONYMOUS_ENABLED=true`,
+`GF_AUTH_ANONYMOUS_ORG_ROLE=Admin`, `GF_AUTH_DISABLE_LOGIN_FORM=true`,
+`GF_AUTH_BASIC_ENABLED=false`), so it ignores the forwarded credentials
+and asks for no second password. `GALAXY_DEV_GRAFANA_ADMIN_PASSWORD`
+still seeds the admin user for provisioning/API use.
+
+Datasources (Prometheus, Loki, Tempo) and a starter dashboard
+(`grafana/dashboards/galaxy-overview.json`) are provisioned as code under
+`grafana/provisioning/`.
+
+## Config delivery
+
+`dev-deploy.yaml` copies this directory to a stable host path
+(`$HOME/.galaxy-dev/monitoring`, exported as `GALAXY_DEV_MONITORING_DIR`)
+before `compose up`, and the compose binds it read-only into the
+collectors. A stable path — not the ephemeral CI workspace — keeps the
+mounts valid across container restarts and host reboots (the same lesson
+as the geoip volume; see `../KNOWN-ISSUES.md`).
+
+## Tuning (cost knobs)
+
+Defaults favour the smallest workable footprint; all are config/compose
+values:
+
+- Prometheus `scrape_interval=30s`, `--storage.tsdb.retention.time=15d`.
+- Loki `retention_period=168h` (7d); Tempo `block_retention=72h` (3d).
+- cAdvisor `--housekeeping_interval=30s`.
+- Per-service `deploy.resources.limits.memory` caps (~1.5 GB total cap;
+  steady-state well under that).
+
+Seven always-on containers cost roughly ~1.1 GB steady RAM and
+~1.5–2.5 GB disk at these retention windows. cAdvisor is the main CPU
+cost; on a constrained host it can be dropped (host + app metrics still
+cover most needs).

tools/dev-deploy/monitoring/grafana/dashboards/galaxy-overview.json

@@ -0,0 +1,46 @@
+{
+  "annotations": { "list": [] },
+  "editable": true,
+  "graphTooltip": 0,
+  "panels": [
+    {
+      "datasource": { "type": "prometheus", "uid": "prometheus" },
+      "fieldConfig": { "defaults": { "unit": "reqps" }, "overrides": [] },
+      "gridPos": { "h": 8, "w": 12, "x": 0, "y": 0 },
+      "id": 1,
+      "title": "Backend HTTP request rate",
+      "type": "timeseries",
+      "targets": [
+        {
+          "datasource": { "type": "prometheus", "uid": "prometheus" },
+          "expr": "sum by (group) (rate(http_requests_total[5m]))",
+          "legendFormat": "{{group}}"
+        }
+      ]
+    },
+    {
+      "datasource": { "type": "prometheus", "uid": "prometheus" },
+      "fieldConfig": { "defaults": { "unit": "bytes" }, "overrides": [] },
+      "gridPos": { "h": 8, "w": 12, "x": 12, "y": 0 },
+      "id": 2,
+      "title": "Container memory (cadvisor)",
+      "type": "timeseries",
+      "targets": [
+        {
+          "datasource": { "type": "prometheus", "uid": "prometheus" },
+          "expr": "sum by (name) (container_memory_usage_bytes{name=~\"galaxy-dev-.*|galaxy-game-.*\"})",
+          "legendFormat": "{{name}}"
+        }
+      ]
+    }
+  ],
+  "schemaVersion": 39,
+  "tags": ["galaxy"],
+  "templating": { "list": [] },
+  "time": { "from": "now-6h", "to": "now" },
+  "timepicker": {},
+  "title": "Galaxy — overview",
+  "uid": "galaxy-overview",
+  "version": 1,
+  "weekStart": ""
+}

tools/dev-deploy/monitoring/grafana/provisioning/dashboards/dashboards.yml

@@ -0,0 +1,12 @@
+# Grafana dashboard provider: load every JSON under the mounted
+# dashboards directory at startup (provisioned as code).
+apiVersion: 1
+
+providers:
+  - name: galaxy
+    type: file
+    disableDeletion: false
+    allowUiUpdates: true
+    options:
+      path: /var/lib/grafana/dashboards
+      foldersFromFilesStructure: false

tools/dev-deploy/monitoring/grafana/provisioning/datasources/datasources.yml

@@ -0,0 +1,22 @@
+# Grafana datasources provisioned as code (dev↔prod parity). All reach
+# the collectors by Docker DNS (compose service names) on
+# galaxy-dev-internal.
+apiVersion: 1
+
+datasources:
+  - name: Prometheus
+    type: prometheus
+    access: proxy
+    uid: prometheus
+    url: http://galaxy-prometheus:9090
+    isDefault: true
+  - name: Loki
+    type: loki
+    access: proxy
+    uid: loki
+    url: http://galaxy-loki:3100
+  - name: Tempo
+    type: tempo
+    access: proxy
+    uid: tempo
+    url: http://galaxy-tempo:3200

tools/dev-deploy/monitoring/loki/loki.yml

@@ -0,0 +1,47 @@
+# Single-binary Loki for the dev stack: filesystem storage, in-memory
+# ring, 7-day retention. Internal-only (no host port).
+auth_enabled: false
+
+server:
+  http_listen_port: 3100
+  grpc_listen_port: 9095
+  log_level: warn
+
+common:
+  instance_addr: 127.0.0.1
+  path_prefix: /loki
+  storage:
+    filesystem:
+      chunks_directory: /loki/chunks
+      rules_directory: /loki/rules
+  replication_factor: 1
+  ring:
+    kvstore:
+      store: inmemory
+
+schema_config:
+  configs:
+    - from: 2024-01-01
+      store: tsdb
+      object_store: filesystem
+      schema: v13
+      index:
+        prefix: index_
+        period: 24h
+
+limits_config:
+  retention_period: 168h
+  reject_old_samples: true
+  reject_old_samples_max_age: 168h
+
+compactor:
+  working_directory: /loki/compactor
+  retention_enabled: true
+  delete_request_store: filesystem
+
+query_range:
+  results_cache:
+    cache:
+      embedded_cache:
+        enabled: true
+        max_size_mb: 64

tools/dev-deploy/monitoring/prometheus/prometheus.yml

@@ -0,0 +1,24 @@
+# Prometheus scrape config for the dev observability stack. Retention is
+# a CLI flag in the compose command, not here. Targets are reached by
+# Docker DNS (compose service names) on galaxy-dev-internal; nothing is
+# published to the host.
+global:
+  scrape_interval: 30s
+  evaluation_interval: 30s
+
+scrape_configs:
+  - job_name: backend
+    static_configs:
+      - targets: ["galaxy-backend:9100"]
+  - job_name: gateway
+    static_configs:
+      - targets: ["galaxy-api:9191"]
+  - job_name: node
+    static_configs:
+      - targets: ["galaxy-node-exporter:9100"]
+  - job_name: cadvisor
+    static_configs:
+      - targets: ["galaxy-cadvisor:8080"]
+  - job_name: prometheus
+    static_configs:
+      - targets: ["localhost:9090"]

tools/dev-deploy/monitoring/promtail/promtail.yml

@@ -0,0 +1,30 @@
+# Promtail tails the dev stack's container logs via the Docker API
+# (service discovery filtered to the galaxy.stack=dev-deploy label) and
+# ships them to Loki. Requires the Docker socket mounted read-only.
+server:
+  http_listen_port: 9080
+  grpc_listen_port: 0
+  log_level: warn
+
+positions:
+  filename: /tmp/positions.yaml
+
+clients:
+  - url: http://galaxy-loki:3100/loki/api/v1/push
+
+scrape_configs:
+  - job_name: docker
+    docker_sd_configs:
+      - host: unix:///var/run/docker.sock
+        refresh_interval: 15s
+        filters:
+          - name: label
+            values: ["galaxy.stack=dev-deploy"]
+    relabel_configs:
+      - source_labels: ["__meta_docker_container_name"]
+        regex: "/?(.*)"
+        target_label: container
+      - source_labels: ["__meta_docker_container_label_galaxy_game_id"]
+        target_label: game_id
+      - source_labels: ["__meta_docker_container_log_stream"]
+        target_label: stream

tools/dev-deploy/monitoring/tempo/tempo.yml

@@ -0,0 +1,30 @@
+# Single-binary Tempo for the dev stack: OTLP receivers, local block
+# storage, 3-day retention. Internal-only (no host port). Backend and
+# gateway push traces here over OTLP gRPC (4317).
+server:
+  http_listen_port: 3200
+  log_level: warn
+
+distributor:
+  receivers:
+    otlp:
+      protocols:
+        grpc:
+          endpoint: 0.0.0.0:4317
+        http:
+          endpoint: 0.0.0.0:4318
+
+ingester:
+  max_block_duration: 5m
+
+compactor:
+  compaction:
+    block_retention: 72h
+
+storage:
+  trace:
+    backend: local
+    local:
+      path: /var/tempo/blocks
+    wal:
+      path: /var/tempo/wal

tools/local-dev/Makefile

@@ -22,7 +22,7 @@ help:
 	@echo "  make up             Build (if needed) and bring up the stack, wait until healthy"
 	@echo "  make down           Stop compose containers, leave engines + volumes intact"
 	@echo "  make rebuild        Force rebuild of backend / gateway images and bring up"
-	@echo "  make build-engine   Build the engine image $(ENGINE_IMAGE) used by the dev sandbox"
+	@echo "  make build-engine   Build the engine image $(ENGINE_IMAGE) used by running games"
 	@echo "  make stop-engines   Stop and remove only the per-game engine containers"
 	@echo "  make prune-broken-engines  Remove non-running engine containers Docker can't heal (run inside 'up')"
 	@echo "  make clean          Stop everything (incl. engines) and wipe volumes + game state"
@@ -37,8 +37,9 @@ help:
 	@echo "  pnpm -C ui/frontend dev"
 	@echo "and open http://localhost:5173 (UI) plus http://localhost:8025 (Mailpit)."
 	@echo ""
-	@echo "Default login for the auto-provisioned dev sandbox: dev@local.test"
+	@echo "Sign in with email-OTP; the fixed login code 123456 works when"
-	@echo "(see BACKEND_DEV_SANDBOX_EMAIL in .env). Login code: 123456."
+	@echo "BACKEND_AUTH_DEV_FIXED_CODE is set in .env. No game is auto-provisioned —"
+	@echo "load a legacy report via the UI's DEV report loader to exercise the map."
 
 up: build-engine prune-broken-engines
 	$(COMPOSE) up -d --wait
@@ -88,12 +89,9 @@ stop-engines:
 # bind-mount source and leaves it stuck in `exited` / `created`
 # state. This target prunes the husks before `compose up`; the
 # backend's pre-bootstrap reconciler tick (`backend/cmd/backend/main.go`)
-# then cascades the orphan runtime row to `removed`, the lobby
+# then cascades the orphan runtime row to `removed` and the lobby
-# cancels the game, and the dev-sandbox bootstrap purges the
+# cancels the game. Healthy `running` / `restarting` containers are
-# cancelled tile and provisions a fresh sandbox in the same
+# left intact so a long-lived game survives normal up/down cycles.
-# `make up` cycle. Healthy `running` / `restarting` containers are
-# left intact so a long-lived sandbox survives normal up/down
-# cycles.
 prune-broken-engines:
 	@ids=""; \
 	for cid in $$(docker ps -aq \

tools/local-dev/README.md

@@ -78,49 +78,24 @@ To force the second path (no fast-bypass), edit
 `make rebuild` (or simply `docker compose up -d backend` to recreate
 the backend with the new env).
 
-## Auto-provisioned dev sandbox
+## No auto-provisioned game
 
-`make up` provisions a private game called **Dev Sandbox** owned by
+`make up` brings up the stack with an empty lobby — there is no
-the dev user (default `dev@local.test`). The flow is implemented in
+auto-provisioned game. Sign in with email-OTP (the fixed dev code
-`backend/internal/devsandbox` and runs on every backend boot when
+`123456` works when `BACKEND_AUTH_DEV_FIXED_CODE` is set in
-`BACKEND_DEV_SANDBOX_EMAIL` is non-empty in `tools/local-dev/.env`.
+`tools/local-dev/.env`):
-
-Bootstrap is idempotent — re-running `make up` after a `make down`
-finds the existing user, dummy participants, game, and memberships
-without creating duplicates. If a previous boot crashed mid-way
-(game stuck in `enrollment_open` or `ready_to_start`), the next boot
-resumes the lifecycle.
-
-To log in straight into the sandbox:
 
 1. `make -C tools/local-dev up`
 2. `pnpm -C ui/frontend dev` (in another terminal)
-3. Open <http://localhost:5173/login>, enter `dev@local.test`, then
+3. Open <http://localhost:5173/login>, enter your email, then the dev
-   the dev code `123456`.
+   code `123456`.
-4. The lobby shows **Dev Sandbox** in *My Games*; click in.
 
-To disable the bootstrap, clear `BACKEND_DEV_SANDBOX_EMAIL` in
+To exercise the map and report views without running a full game, use
-`tools/local-dev/.env` and `docker compose up -d backend` (or
+the UI's DEV **synthetic report loader**: convert a legacy `.REP` with
-`make rebuild`). Existing users / games are not removed.
+`tools/local-dev/legacy-report/` and load the resulting JSON through the
-
+loader (see that tool's README). To play a real game, create one in the
-Terminal sandbox games — anything in `cancelled`, `finished`, or
+lobby and let the engine (`galaxy-engine:local-dev`, built by
-`start_failed` — are deleted on every boot before find-or-create
+`make build-engine`) run it.
-runs. The cascade declared in `00001_init.sql` removes the
-matching memberships, applications, invites, runtime records,
-and player mappings in the same write, so the dev user's lobby
-shows exactly one running tile at all times. Cancelling the
-sandbox manually and running `docker compose restart backend`
-(or `make rebuild`) yields a fresh game without leaving dead
-tiles behind.
-
-The bootstrap requires:
-- `galaxy-engine:local-dev` Docker image (`make build-engine`).
-- `BACKEND_DEV_SANDBOX_ENGINE_VERSION` parses as plain semver
-  (`MAJOR.MINOR.PATCH`); the default `0.1.0` is what the bootstrap
-  registers in the `engine_versions` row that points at the image.
-- `BACKEND_DEV_SANDBOX_PLAYER_COUNT` ≥ 20 (the engine's minimum;
-  19 deterministic dummies fill the slots so the single real user
-  can start the game).
 - A frozen turn schedule (`0 0 1 1 *` — once a year) so the visible
   game state stays at turn 1 until you explicitly progress it.
 
@@ -239,24 +214,15 @@ make status         docker compose ps
   this in one cycle: `prune-broken-engines` (runs as part of `up`)
   removes every engine container that is not in `running` /
   `restarting` state, the backend's pre-bootstrap reconciler tick
-  cascades the orphan runtime row to `removed`, the lobby cancels
+  cascades the orphan runtime row to `removed`, and the lobby cancels
-  the matching sandbox game, and the dev-sandbox bootstrap purges
+  the matching game. To run the cleanup by hand without restarting the
-  the cancelled tile and provisions a fresh sandbox with a brand
+  rest of the stack, `make prune-broken-engines`.
-  new state directory. To run the cleanup by hand without restarting
-  the rest of the stack, `make prune-broken-engines`.
 
   The cycle relies on the backend image carrying the pre-bootstrap
   reconciler tick (`backend/cmd/backend/main.go`). `make up` reuses
   the cached image, so after pulling this commit the first time you
   must `make rebuild` once to bake the fix in. Future `make up`
   cycles will heal in one shot.
-
-  If after the heal cycle the lobby still shows only a `cancelled`
-  sandbox tile and no running game, the running backend image
-  predates the pre-bootstrap reconciler tick — the periodic ticker
-  cancels the orphan after bootstrap has already returned, leaving
-  the lobby in the half-baked state. `make rebuild` recreates the
-  image and then `make up` lands a fresh sandbox.
 - **`make up` reports a build error mentioning `pkg/cronutil`** —
   upstream module list drifted; copy any new `pkg/<name>/` line into
   the local-dev `backend.Dockerfile` / `gateway.Dockerfile` to match

tools/local-dev/docker-compose.yml

@@ -122,10 +122,6 @@ services:
       BACKEND_OTEL_TRACES_EXPORTER: none
       BACKEND_OTEL_METRICS_EXPORTER: none
       BACKEND_AUTH_DEV_FIXED_CODE: ${BACKEND_AUTH_DEV_FIXED_CODE:-}
-      BACKEND_DEV_SANDBOX_EMAIL: ${BACKEND_DEV_SANDBOX_EMAIL:-}
-      BACKEND_DEV_SANDBOX_ENGINE_IMAGE: ${BACKEND_DEV_SANDBOX_ENGINE_IMAGE:-}
-      BACKEND_DEV_SANDBOX_ENGINE_VERSION: ${BACKEND_DEV_SANDBOX_ENGINE_VERSION:-}
-      BACKEND_DEV_SANDBOX_PLAYER_COUNT: ${BACKEND_DEV_SANDBOX_PLAYER_COUNT:-}
     volumes:
       - /var/run/docker.sock:/var/run/docker.sock
       # Per-game state directories live under the same absolute path

ui/docs/game-state.md

@@ -85,16 +85,16 @@ report to fetch. Two alternatives were rejected:
 - a brand-new `user.games.state` message — adds a full wire-flow
   (fbs schema, transcoder, gateway routing, backend handler) for a
   one-field response;
-- hard-coding `turn=0` for all games — works for the dev sandbox
+- hard-coding `turn=0` for all games — works for a synthetic report
-  (which never advances past turn zero) but renders the initial
+  loaded at turn zero but mis-renders the initial state for any real
-  state for any real game past turn zero.
+  game past turn zero.
 
 Extending `GameSummary` reuses the existing lobby pipeline; the
 backend already tracks `current_turn` in its runtime projection
 (`backend/internal/server/handlers_user_lobby_helpers.go`
 `gameSummaryToWire` reads it from `g.RuntimeSnapshot.CurrentTurn`).
 The `current_turn` field defaults to zero on the FB side, so existing
-tests and the dev sandbox flow continue to work unchanged.
+tests and the synthetic-report flow continue to work unchanged.
 
 ## State binding
 

ui/frontend/src/api/synthetic-report.ts

@@ -1,6 +1,8 @@
 // DEV-only synthetic-report loader. Backs the "Load synthetic report"
-// affordance on the lobby (visible behind `import.meta.env.DEV`) and
+// affordance on the lobby (visible when the build-time flag
-// the in-game shell layout's bypass for the synthetic game id range.
+// `VITE_GALAXY_DEV_AFFORDANCES === "true"` — the dev and dev-deploy
+// bundles; stripped from prod) and the in-game shell layout's bypass
+// for the synthetic game id range.
 //
 // The accepted JSON shape mirrors `pkg/model/report.Report` as
 // emitted by `tools/local-dev/legacy-report/cmd/legacy-report-to-json`.