docs: update open threads with Feb 22 state

OPEN_THREADS.md

@@ -11,38 +11,66 @@ This document tracks **known open questions and deferred decisions**.
 - ✅ Customer isolation (API + frontend)
 - ✅ Agent update system (versioned, hash-verified)
 - ✅ Minecraft player presence (agent-sourced)
-- ✅ Forge startup race condition
+- ✅ Game telemetry router separation (`/api/game/*`)
-- ✅ Prometheus SD schema alignment
+- ✅ Agent Phase 1 mod management endpoints
+- ✅ Agent process metrics endpoint
+- ✅ Minecraft readiness probe + restart race mitigation (initial)
 
 ---
 
-## Still Open
+## Still Open (Phase 1 / Go-to-market)
 
-### Console UI (Final Polish)
+### API: proxy the new agent capabilities
-- Reduce remaining visual noise in game console
+- Add game-mod endpoints behind auth + ownership:
-- Confirm final placement/behavior of player list (inline vs overlay)
+  - `GET    /api/game/servers/:vmid/mods`
-- Lock icon sizing + metric density
+  - `POST   /api/game/servers/:vmid/mods/install`
-- Final accessibility contrast pass
+  - `PATCH  /api/game/servers/:vmid/mods/:mod_id`
+  - `DELETE /api/game/servers/:vmid/mods/:mod_id`
+- Add process metrics proxy:
+  - `GET /api/game/servers/:vmid/metrics/process`
+- Decide caching shape (Redis keys + TTLs) for:
+  - mods list (agent already caches 5m internally)
+  - process metrics (likely low TTL, e.g. 5–10s)
 
-### Metrics & Observability
+### Artifacts: Modrinth / NeoForge supply-chain workflow
-- Decide on caching strategy for metrics API (Redis vs in-process)
+- Define a **local-first** artifact ingestion workflow:
-- Determine long-term retention strategy (Prometheus vs downsampling)
+  - detect → pending → verified → active → deprecated
-- Admin-only Grafana dashboards (internal)
+- Decide Phase 1 scope:
+  - curated mod list vs modpack-first
+  - whether to ingest only Modrinth + NeoForge initially
+- Standardize loader naming contracts:
+  - `minecraft/neoforge/<mc_version>/neoforge-installer.jar` (Phase 1)
+  - confirm similar conventions for Forge/Fabric installers
 
-### Provisioning & Scale
+### Frontend: Mods UI (Phase 1)
+- Add "Mods" surface for game servers:
+  - installed mods list (enabled/disabled)
+  - enable/disable toggle
+  - remove mod
+  - curated install flow (no arbitrary URLs)
+- Decide if "upload custom mods" is Phase 1 or Phase 2:
+  - Phase 2 is preferred unless there is sandboxing/scanning/allowlist.
+
+### Provisioning & scale validation
 - Validate concurrent provisioning (multi-user, parallel creates)
-- Stress-test agent update under load
+- Stress-test agent update checks under load (notify/auto modes)
-- Define safe limits for agent restart frequency
+- Confirm safe limits for restart/backoff behavior
 
-### Agent Hardening
+---
-- Rollback behavior if new agent fails health checks
-- Optional manifest signing (GPG / minisign)
-- Upgrade window coordination (future)
 
-### Game-Specific Enhancements (Deferred)
+## Still Open (Polish / Phase 2)
-- Mods/runtime badges (Paper / Fabric / Forge)
+
-- World/map visualization (Minecraft-specific, future phase)
+### Console UI
-- Player detail views (separate surface, not console)
+- Finalize player list UX (overlay vs inline)
+- Final accessibility/contrast pass
+
+### Agent hardening
+- Rollback behavior if a new agent fails health checks after update
+- Optional manifest signing (minisign/GPG)
+
+### Game platform roadmap
+- Modpack-first install flow (server files + manifests)
+- Engine abstraction for additional games (Rust/Terraria/Valheim)
 
 ---
 
@@ -52,3 +80,4 @@ This document tracks **known open questions and deferred decisions**.
 - Live chat overlays
 - Grafana iframe embedding for customers
 - Per-player stats in the main console
+- Arbitrary mod uploads without scanning/sandboxing