Skip to main content
Cyberwave is built in the open with a small, responsive team and a growing community. This page is the single place to figure out where to go when you need help — whether you’re stuck on a quickstart, hit a bug, or want to talk to a human.

​
Where to get help

Discord community

The fastest place to ask a question, share what you’re building, or get unstuck. The team and other builders hang out here every day.

GitHub issues

Found a bug or want to request a feature? Open an issue on the relevant repo under the cyberwave-os organization (SDKs, CLIs, drivers, edge components, this docs site).

Contact us

For sales, partnerships, on-premises deployments, or anything that doesn’t belong in a public channel, reach out directly.

Platform status

Live uptime for the API, MQTT broker, and app. Check here first — if something is down, it’s probably not your code.

Changelog

Check what shipped recently and whether a recent change might explain what you’re seeing.
Not sure which channel to use? Default to Discord for questions and GitHub issues for reproducible bugs or feature requests.

​
Filing a good GitHub issue

A reproducible report gets fixed dramatically faster. When opening an issue on a cyberwave-os repo, please include:
  • What you ran: the command, code snippet, or UI action.
  • What you expected vs. what happened (include the full error or log output).
  • Environment: OS, Python / Node version, SDK or CLI version, and whether you’re on cloud or self-hosted.
  • Catalog item (if relevant): the asset, model, or driver slug — e.g. the-robot-studio/so101, unitree/go2.
  • A minimal reproduction if you can — even a 10-line script helps.
Pick the repo closest to where the issue lives (Python SDK, CLI, a specific edge driver, the docs, etc.). If you’re not sure, ask on Discord first and we’ll point you at the right place.

​
Catalog-item help (robots, models, datasets)

Most “how do I…” or “why isn’t this working” questions about a specific catalog item are answered on that item’s own page rather than here. Every catalog entry has its own setup guide, capabilities reference, and item-specific troubleshooting. Start there before opening a general issue:

Robot hardware

Per-robot setup, supported features, calibration tips, and known issues for each supported arm, mobile base, quadruped, and drone.

Model catalog

Per-model usage, supported inputs/outputs, training notes, and gotchas for each model in the catalog.

Asset catalog

How catalog assets, registry IDs, and slugs work — and how to upload your own.

Datasets & recordings

How datasets are generated from recordings, supported formats, and import/export options.
Rule of thumb: if your question starts with “for the SO-101…” / “for Go2…” / “for SmolVLA…”, the answer is most likely on that item’s page. This Support page covers cross-cutting questions about the platform itself.

​
FAQs

A growing list of frequently asked questions about the platform as a whole. For item-specific questions, see the catalog links above.

​
General

Cyberwave turns every robot and sensor into a digital twin that your code, your models and your agents can drive, with the same API in simulation and on real hardware. See Connect AI to robots.
A simulator models a robot. A Cyberwave twin is the same object in simulation and on the real robot: cw.affect("simulation") or cw.affect("live") switches which one your code drives. See What is a digital twin and Simulation.
No. Develop in simulation runs in the browser with no code, and Your first AI loop needs only a laptop webcam. To drive a real arm without owning one, see Develop with remote hardware.
Playground simulation is free. MuJoCo simulations, AI usage, platform events and storage use credits. See Credits & Billing.

​
Accounts, workspaces & pricing

A workspace is your team’s container for environments, workflows and resources, with its own members, permissions and billing. See Data model.
Share a project from its settings: click Share, add people by email and pick a role. See Access control.
In credits. Platform events, AI usage and storage are metered, and a MuJoCo simulation costs about 0.6 credits per hour. The Usage page in your organization settings shows rates and charges. See Credits & Billing.
A self-hosted package exists; contact Cyberwave. Air-gapped deployment is not available as a packaged option. See Cyberwave for integrators.

​
Hardware compatibility

See Supported hardware for the compatibility table and the catalog for every device.
Yes. Create an asset from its URDF and write a driver. See Add your robot, custom hardware and writing compatible drivers.
USB, RTSP, GigE and RealSense cameras, plus cameras, microphones, and other sensors.
Yes. Drivers can be ROS 2 nodes built on BaseROS2Driver. See Supported hardware.

​
SDKs & integrations

Create a token under Profile → Access and export it as CYBERWAVE_API_KEY. See API tokens and Authentication.
Yes. Point your client at the hosted MCP server, https://mcp.cyberwave.com/mcp, with your API key as a Bearer token. See the MCP server.
Any language with an HTTP client. Send Authorization: Bearer <api-key> to https://api.cyberwave.com. See the API overview.

​
Edge & deployment

Edge Core runs on the computer next to your robot. It registers the device and runs drivers and workers as Docker containers. See Edge runtime.
Raspberry Pi, Jetson, x86 Linux machines and macOS. See Raspberry Pi and Jetson Orin Nano.
Offline behaviour is not documented yet. Edge nodes need outbound access to api.cyberwave.com and mqtt.cyberwave.com; see Network and firewall requirements, and ask on Discord for details.
Subclass the SDK’s driver base class and declare its interface. See Writing compatible drivers.

​
AI models, workflows & datasets

Yes, in early access: call your own API or hand Cyberwave code to run. See Bring your own model.
On the edge, workers run Ultralytics (.pt), ONNX Runtime, TFLite and OpenCV models, each behind an SDK extra (ml, ml-onnx, ml-tflite, ml-opencv). See Model cache.
Wire triggers, models, conditions and robot commands on the visual canvas, or ask the workflow assistant. See Build and run workflows.
From recordings of your twins, or by importing from Hugging Face Hub or a zip file.

​
Teleoperation & streaming

Latency figures are not published yet. See Live Teleoperation.
Usually, yes. Every connection is outbound, and video relays over TLS on port 443. See Network and firewall requirements.
Not documented yet. Contact us if you need a specific browser.

​
Security & data

Hosting regions and retention are not documented yet. Contact us.
In transit, REST uses HTTPS, MQTT uses TLS on port 8883, and the video relay uses TLS on port 443. For encryption at rest, contact us.
Workspace members, according to their roles, people you share a project with, and anyone with a project’s public link. See access control.
Ask us. Contact us for current certifications.

​
Still stuck?

If none of the above unblocks you:
  1. Check platform status — confirm whether it’s us before you debug it.
  2. Search the cyberwave-os GitHub org for an existing issue.
  3. Ask in Discord — usually the fastest path to an answer.
  4. Open a new GitHub issue on the relevant repo with the details from the filing a good issue checklist above.
  5. For private or commercial questions, contact us directly.