question: Should we rewire to Fluxer for chat+voice+video?
answer: NO — Fluxer is the WRONG tool for embedding into Funday.
why:
  - Fluxer is a Discord-CLONE app (full Electron client + backend)
  - NOT a library/SDK — it's a whole product
  - Heavy stack (Erlang+TS+Rust+Cassandra+Meilisearch+Valkey+LiveKit)
  - Replaces Nakama entirely — destroys our existing chat infrastructure
  - We already have ~691 chat references across 36 frontend files
recommendation: Nakama (keep) + LiveKit (add)
why_this:
  - Fluxer itself uses LiveKit for voice/video → use LiveKit directly
  - Nakama already gives us hierarchical channels (room/group/DM)
  - LiveKit = Go + Pion WebRTC (matches our stack), K3s-native, SFU
  - Bridge them via shared room/channel ID convention

// nakama-modules/livekit_rpc.ts
import { AccessToken } from "livekit-server-sdk"
 
export const rpcLiveKitToken: nkruntime.RpcFunction = (ctx, logger, nk, payload) => {
  const { scope, type, id } = JSON.parse(payload) // e.g. {scope:'match', type:'connect4', id:'abc123'}
  const roomName = `${scope}:${type}:${id}`
 
  // Authorization: verify user has access to this Nakama resource
  const userId = ctx.userId
  if (!userCanAccess(nk, userId, scope, type, id)) {
    throw Error("forbidden")
  }
 
  const at = new AccessToken(LIVEKIT_API_KEY, LIVEKIT_API_SECRET, {
    identity: userId,
    name: ctx.username,
    metadata: JSON.stringify({ scope, type, id }),
  })
  at.addGrant({
    room: roomName,
    roomJoin: true,
    canPublish: true,
    canSubscribe: true,
    canPublishData: false, // Nakama owns text data
  })
  return JSON.stringify({ token: at.toJwt(), url: "wss://livekit.funday.gg", room: roomName })
}

// frontend/src/lib/comms/voice.ts
import { Room, RoomEvent, Track } from "livekit-client"
import { nakama } from "$lib/server/nakama"
 
export async function joinVoice(scope: string, type: string, id: string) {
  const session = await nakama.getSession()
  const { token, url, room } = await nakama.callRpc(session, "livekit_token", { scope, type, id })
 
  const lkRoom = new Room({ adaptiveStream: true, dynacast: true })
  await lkRoom.connect(url, token)
  await lkRoom.localParticipant.setMicrophoneEnabled(true)
  return lkRoom
}

// nakama-modules/match-lifecycle.ts
// On match end → close LiveKit room
async function onMatchEnded(matchId: string) {
  await fetch(`https://livekit.funday.gg/twirp/livekit.RoomService/DeleteRoom`, {
    method: "POST",
    headers: livekitAuthHeaders(),
    body: JSON.stringify({ room: `match:${gameId}:${matchId}` }),
  })
}

# Helm install
helm repo add livekit https://helm.livekit.io
helm install livekit livekit/livekit-server \
  -n livekit --create-namespace \
  --set redis.enabled=true \
  --set livekit.keys.<API_KEY>=<API_SECRET>

# gitops/platform/base/livekit/livekit-ingress.yaml
apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  name: livekit-ingress
  namespace: livekit
  annotations:
    traefik.ingress.kubernetes.io/router.priority: "950"
spec:
  tls:
    - hosts: [funday.gg]
      secretName: funday-tls-cert
  rules:
    - host: funday.gg
      http:
        paths:
          - path: /livekit
            pathType: Prefix
            backend: { service: { name: livekit, port: { number: 7880 } } }

# UDP 50000-60000 for media
ufw allow 50000:60000/udp
# TCP 7881 for fallback
ufw allow 7881/tcp

pnpm add livekit-client
# Add /lib/comms/voice.ts + /lib/comms/video.ts wrappers

pnpm add livekit-server-sdk
# Add nakama-modules/livekit_rpc.ts
# Register in index.ts: initializer.registerRpc('livekit_token', rpcLiveKitToken);

fluxer: NO — wrong abstraction level (whole app, not embeddable)
recommendation: Nakama (keep all 691 refs) + LiveKit (add for voice/video)
hierarchy_bridge: shared <scope>:<type>:<id> naming convention
deployment: 1 K3s helm chart + 1 RPC + 2 frontend modules
effort: ~1 sprint to MVP, ~2 sprints with screen-share + recording
risk: low — both stacks are battle-tested in our environment