Common Questions

No. Sharkord covers core chat and voice for small and medium private groups, with plugins for everything else. It is not trying to replicate every Discord feature, and it is not built for large public communities.

The server is light on CPU and memory; bandwidth is what you actually plan for. Voice and video are routed, not mixed, so a channel with N people talking means the server forwards each stream to N-1 others. Screen sharing at a high bitrate dominates everything else, and webRtc.maxBitrate in config.ini is the knob that caps it.

A small VPS handles a handful of concurrent voice users comfortably. Text-only usage is negligible.

Yes. Stop it, copy the data directory across, and start the same version there. Everything comes with it.

No. One instance is one server. You can run several instances on the same machine with different ports and different data directories, but they stay completely separate.

Uploads are stored on disk, inside the data directory. There is no support for object storage such as S3. You can move the whole data directory, including uploads, with SHARKORD_DATA_PATH, so a mounted volume works.

They are encrypted in transit when you serve Sharkord over HTTPS, and stored unencrypted in the database on your server. There is no end-to-end encryption.

No.

Yes. WebRTC media is always encrypted between the browser and the server with DTLS-SRTP, whether or not you use HTTPS. It is decrypted at the server to be routed, so this is transport encryption, not end-to-end.

Yes. Sharkord does not terminate TLS itself, so put a reverse proxy in front of it. See the HTTPS guides. If you do not own a domain, claim a free subdomain.

Yes, through plugins. There is a small music bot in the marketplace, and you can write your own.

Not yet. Sharkord runs in the browser, including mobile browsers. Native clients are a possibility for the future.

There is no telemetry or analytics, and nothing about your server or its users is reported anywhere. The server does make outbound requests in four cases:

• To look up its own public IP at startup, unless you set webRtc.announcedAddress.
• To GitHub, if you enable automatic updates or open the update panel.
• To the plugin marketplace, when you browse or install plugins.
• To ipinfo.io, when a user logs in, to record the city and country shown in their login history. The user's IP address is sent to that service, and results are cached.

No. AI is used as a tool for development, documentation, and testing, but design and implementation decisions are made by the developers.

Delete the data directory. That removes every user, message, and file, and the next start generates a new server and a new owner token.