Documentation / API reference

Files & upstream

This node's Hyperdrive over HTTP, and the one door through which a prompt can leave the P2P network entirely.

List files

GET/v1/filesBearerTry it →

Lists this node's Hyperdrive. Pass ?peerKey=<hex> to list a peer's drive instead — without it you always get the local one, whichever node you meant.

Upload and fetch

POST/v1/files/upload?name=<name>Bearer
POST/v1/files/fetchBearer

Upload sends the raw body as the file content. Fetch downloads a qvac:// link to this node's disk — the API equivalent of pyrusllm fetch.

Both write to this node's disk, so they generate a curl command in the playground rather than executing — see what the playground restricts.

External assistant

GET/v1/upstreamBearerTry it →

Lists the configured upstreams and whether each has its credential. Configuring a new upstream or changing a credential is not possible over HTTP — that lives in upstreams.json on the operator's disk.

Opt in or out

POST/v1/upstream/opt-inBearer
{ "enabled": true }
This is the one path where a prompt leaves the P2P network and goes to a company's API, which sees it, may log it, and bills it. Off by default. Even switched on it only competes when nobody local or on the network has capacity, and never when the request asks to stay on this machine.
Turning it on here does not write the config file: on restart it goes back to whatever upstreams.json says.