Content Extend is an OpenTechnologyApp project template that replaces two desktop apps — Stream Extend (broadcast control) and Creator Extend (AI content creation) — with one web dashboard connected to the tools actually running on your machines: OBS, ComfyUI, Blender, Krita, Ollama, NodeCG, and the MinIO/Postgres/Redis backing stack.
The connection is a small bridge — a single Node script per machine. Outbound-only: it listens to your project for control events and POSTs telemetry back. No inbound ports, no firewall holes, and your OBS/ComfyUI passwords never leave the machine.
The template directory is going open source on GitHub — watch the org for the repo.
What replaces what
| Extend app surface | Content Extend |
|---|---|
| Tool auto-detect + one-click setup | Tool & Service Inventory queue + per-platform install playbooks |
| Connection status, FPS/bitrate, GPU/VRAM | Health Checks queue — every metric is a tracked item with thresholds |
| Scene switch, go-live, queue workflow | Control Commands fired over webhooks → bridge → local APIs |
| Render queue | Content Pipeline queue, grouped by stage, failures auto-escalated |
| (the apps had nothing) | Incidents queue with post-mortem playbook |
| Cloud relay subscription | The app is the relay — remote access is inherent |
Checks in three tiers
- T1 — Broadcast: OBS heartbeat, stream vitals (FPS / bitrate / dropped frames, evaluated only while live), audio pipeline, broadcast graphics
- T2 — Pipeline: ComfyUI queue depth + VRAM headroom, render-job failures, LLM latency for the AI co-host
- T3 — Infra: Docker backing services, disk/asset storage, tool version drift
Plus the meta-check every distributed system needs: bridge connectivity. If every check goes silent at once, the bridge is down — not your fleet. The template never mass-flags on bridge silence.
The automation that pays for the whole thing
Go Live → pause ComfyUI. When OBS flips to Live, the bridge pauses the render queue sharing the streaming GPU; when the stream ends, it resumes. Encode headroom beats render throughput while you're live — this one rule ends the dropped-frames-during-a-batch-render incident forever. It ships alongside down-service escalation, auto-restart via bridge, failed-render escalation, and Slack incident notification.
Playbooks, not tribal knowledge
- Bridge setup — connect a machine in five checked steps, ending with a control round-trip test
- Go-Live checklist — graphics up, audio sane, GPU gated, vitals watched for two minutes
- Service-down triage — cross-channel confirmation before you touch anything
- Stream degraded / render failed — the diagnostic ladders, in order
- Incident post-mortem — every P1/P2 makes the checks smarter
What stays local, on purpose
Realtime AI effects, capture, encode, and GPU inference never round-trip a server — the dashboard tracks and controls them, it doesn't proxy frames. Silent one-click installs and mobile RTMP broadcasting remain native-app territory; the bridge covers control, telemetry, and guided installs, and is honest about the rest.
Get it
Import the template into OpenTechnologyApp (New Project → Import Template), run the Bridge Setup playbook on each machine, and enable the webhook automations after one trusted cycle. Template JSON, playbooks, and the bridge script are heading to the OpenTechnology14 GitHub org.