Update OPEN_THREADS 2026-03-17 — base-path blocking issue, agent enhancements completed

OPEN_THREADS.md

@@ -15,6 +15,7 @@ Completed:
 - catalog validation implemented
 - runtime installs artifact-backed
 - install guard implemented
+- all installs now fetch from artifact server (no local artifact assumption)
 
 Outstanding:
 
@@ -31,6 +32,7 @@ Completed:
 - dev user creation
 - workspace root `/home/dev/workspace`
 - console runs as dev user
+- `HOME`, `USER`, `LOGNAME`, `TERM` env vars set correctly
 
 Outstanding:
 
@@ -40,64 +42,85 @@ Outstanding:
 
 ---
 
-## Code Server Addon
+### Code Server Addon
 
-Status: ✅ Installed and running inside dev containers
+Status: ✅ Installed and running
 
 Confirmed:
 
-- compiled release artifact fixed on `zlh-artifacts`
-- install confirmed working
-- process confirmed running inside container
+- pulled from artifact server (tar.gz)
+- installed to `/opt/zlh/services/code-server`
 - binds to `0.0.0.0:6000`
-- launched from `/opt/zlh/services/code-server`
+- lifecycle endpoints: `POST /dev/codeserver/start|stop|restart`
+- detection via `/proc/*/cmdline` scan (no longer relies solely on PID file)
 
-Port: `6000`
+**BLOCKING — next task:**
 
-**Next session — agent change required:**
-
-code-server must be relaunched with:
+code-server must launch with:
 
 ```
+--bind-addr 0.0.0.0:6000
 --auth none
+--disable-telemetry
 --base-path /api/dev/<vmid>/ide
+/home/dev/workspace
 ```
 
-Reason: API token is now the sole auth mechanism. Password prompt must be removed. Base path required for correct asset loading through proxy.
+Without `--base-path`, WebSocket paths and static assets mismatch through
+the proxy. Result: IDE loads partially, WS closes with 1006, workspace
+shows `!` (not mounted), extension host fails to start.
+
+---
+
+### Game Server Supervision
+
+Completed:
+
+- crash recovery with backoff: 30s → 60s → 120s
+- backoff resets if uptime ≥ 30s
+- transitions to `error` state after repeated failures
+- crash observability: time, exit code, signal, uptime, log tail, classification
+- classifications: `oom`, `mod_or_plugin_error`, `missing_dependency`, `nonzero_exit`, `unexpected_exit`
+
+---
+
+### Agent Future Work (priority order)
+
+1. **Fix code-server `--base-path` launch arg** — unblocks IDE (IMMEDIATE)
+2. Structured logging (slog) for Loki
+3. Dev container `provisioningComplete` state in `/status`
+4. Graceful shutdown verification (SIGTERM + wait for Minecraft)
+5. Process reattachment on agent restart
 
 ---
 
 ## Dev IDE Access
 
-### Browser IDE (Implemented ✅)
+### Browser IDE
 
 ```
-Browser
-  ↓
-Portal
-  ↓
-API (/api/dev/:id/ide)
-  ↓
-container:6000
+Browser → Portal → API (/api/dev/:id/ide) → container:6000
 ```
 
-Implemented in API:
+API layer: ✅ complete
+Agent layer: ⚠️ blocked on `--base-path`
 
-- `src/routes/devProxy.js` — proxy route mounted in `src/app.js`
-- `GET /api/dev/:id/ide` and `GET /api/dev/:id/ide/*`
-- ownership verification before proxying
-- `ctype === "dev"` required
-- WebSocket support via `http-proxy-middleware` (`ws: true`)
-- `server.on('upgrade')` handler wired
+What is confirmed working:
 
-IDE token system implemented:
+- API auth ✅
+- Token flow ✅
+- Proxy routing ✅
+- WebSocket upgrade handler ✅
+- Upstream targeting ✅
+- code-server process running ✅
 
-- `POST /api/dev/:id/ide-token` — returns signed short-lived token
-- token payload: `sub`, `vmid`, `type: "dev-ide"`
-- default TTL: 300 seconds
-- env overrides: `API_AUTH_IDE_TTL_SECONDS`, `API_AUTH_IDE_SECRET`
-- proxy accepts `Authorization: Bearer` or `?token=<ide-token>`
-- WebSocket upgrades validate same token
+What is failing:
+
+- Workbench WebSocket session ❌
+- Filesystem provider initialization ❌
+- Extension host startup ❌
+
+Root cause: code-server launched without `--base-path /api/dev/<vmid>/ide`
 
 ### Local Dev Access (Headscale/Tailscale — Future)
 
@@ -108,22 +131,7 @@ Outstanding:
 - API auth key generation
 - portal setup instructions
 
-Constraints:
-
-- `magic_dns: false`
-- no exit nodes
-- no DNS takeover
-
----
-
-## Agent Future Work (priority order)
-
-1. Update code-server launch args (`--auth none`, `--base-path /api/dev/<vmid>/ide`)
-2. Structured logging (slog) for Loki
-3. Dev container provisioningComplete state
-4. Crash recovery backoff
-5. Graceful shutdown verification
-6. Process reattachment on agent restart
+Constraints: `magic_dns: false`, no exit nodes, no DNS takeover
 
 ---
 
@@ -136,7 +144,7 @@ Completed:
 - enable_code_server flag
 - `GET /api/servers/:id/status` — server status endpoint
 - `POST /api/dev/:id/ide-token` — IDE token generation
-- `GET /api/dev/:id/ide` — IDE proxy route with WebSocket support
+- `GET /api/dev/:id/ide` + `GET /api/dev/:id/ide/*` — IDE proxy with WebSocket
 - dev routing experiment removed (`devRouting.js`, `devDePublisher.js` deleted)
 
 Outstanding:
@@ -183,3 +191,7 @@ Future work:
 - ✅ API status endpoint for frontend agent-state consumption
 - ✅ Dev IDE proxy implementation (API proxy + token system)
 - ✅ Dev DNS/Traefik routing experiment — removed
+- ✅ Game server crash recovery with backoff
+- ✅ Crash observability (classification, log tail, exit metadata)
+- ✅ Code-server lifecycle endpoints (start/stop/restart)
+- ✅ Code-server process detection via /proc scan