npm package discovery and stats viewer.

Discover Tips

  • General search

    [free text search, go nuts!]

  • Package details

    pkg:[package-name]

  • User packages

    @[username]

Sponsor

Optimize Toolset

I’ve always been into building performant and accessible sites, but lately I’ve been taking it extremely seriously. So much so that I’ve been building a tool to help me optimize and monitor the sites that I build to make sure that I’m making an attempt to offer the best experience to those who visit them. If you’re into performant, accessible and SEO friendly sites, you might like it too! You can check it out at Optimize Toolset.

About

Hi, 👋, I’m Ryan Hefner  and I built this site for me, and you! The goal of this site was to provide an easy way for me to check the stats on my npm packages, both for prioritizing issues and updates, and to give me a little kick in the pants to keep up on stuff.

As I was building it, I realized that I was actually using the tool to build the tool, and figured I might as well put this out there and hopefully others will find it to be a fast and useful way to search and browse npm packages as I have.

If you’re interested in other things I’m working on, follow me on Twitter or check out the open source projects I’ve been publishing on GitHub.

I am also working on a Twitter bot for this site to tweet the most popular, newest, random packages from npm. Please follow that account now and it will start sending out packages soon–ish.

Open Software & Tools

This site wouldn’t be possible without the immense generosity and tireless efforts from the people who make contributions to the world and share their work via open source initiatives. Thank you 🙏

© 2026 – Pkg Stats / Ryan Hefner

@sobree/collab-server

v0.1.13

Published

Node-only Y-protocol relay + persister for Sobree. WebSocket fan-out, filesystem snapshots, pluggable auth.

Readme

@sobree/collab-server

Node-only y-protocol relay + persister for Sobree.

→ Docs: docs.sobree.dev/api/collab-server

One process can host many rooms; each room is one Y.Doc shared by N peers. The server speaks pure y-protocol — it doesn't instantiate an Editor and doesn't know about Sobree's AST or DOCX format. This is the load-bearing simplification that lets one Node process host thousands of rooms cheaply.

Install

pnpm add @sobree/collab-server

Quick start

import { SobreeCollabServer, filesystemPersistence } from "@sobree/collab-server";

const server = new SobreeCollabServer({
  port: 1234,
  persistence: filesystemPersistence({ dir: "./data" }),
});

await server.listen();
console.log("collab server on ws://localhost:1234/<room-id>");

Clients connect via the standard y-websocket provider — the server speaks the same wire format. From a Sobree app:

import * as Y from "yjs";
import { createSobree } from "@sobree/core";
import { attachWebsocketProvider } from "@sobree/collab-providers";

const ydoc = new Y.Doc();
const collab = await attachWebsocketProvider(ydoc, {
  url: "ws://localhost:1234",
  room: "doc-q2-brief",
  name: "Alice",
  color: "#f59e0b",
});
await collab.synced;
const editor = createSobree("#editor", { ydoc });

Persistence

Two backends ship out of the box:

import { filesystemPersistence, memoryPersistence } from "@sobree/collab-server";

// One file per room under ./data/<room-path>.ydoc
filesystemPersistence({ dir: "./data" });

// Ephemeral; lost on restart. Useful for tests / dev.
memoryPersistence();

For production at scale, write a custom backend against S3 / R2 / Postgres / Redis — the interface is small:

interface Persistence {
  load(roomId: string): Promise<Uint8Array | null>;
  save(roomId: string, update: Uint8Array): Promise<void>;
  delete?(roomId: string): Promise<void>;
}

Pass any conformant object to SobreeCollabServer via persistence.

Auth

Optional onConnection hook runs before the peer joins a room. Three return shapes:

  • true — accept with full read/write
  • false — reject (WebSocket closes with code 1008)
  • { allow: true, write: false } — accept as read-only
const server = new SobreeCollabServer({
  port: 1234,
  persistence: filesystemPersistence({ dir: "./data" }),
  onConnection: async ({ req, roomId }) => {
    const token = new URL(req.url ?? "", "http://x").searchParams.get("token");
    const claims = await verifyToken(token, roomId);
    if (!claims) return false;
    return { allow: true, write: claims.role !== "viewer" };
  },
});

The hook receives the raw HTTP upgrade request, so headers / cookies / client IP are all available.

Read-only peers

Read-only peers can read and publish presence (cursors) but their Y sync-update messages are dropped server-side. They see other peers' edits arrive normally.

onConnection: ({ roomId }) => {
  if (roomId.startsWith("public/")) {
    return { allow: true, write: false };
  }
  return { allow: true, write: true };
}

Leader-election for empty-room seeding

When a peer joins a room, the server sends a session message (message type 2, JSON-encoded) immediately, before any sync state:

interface SessionPayload {
  isEmpty: boolean;     // true if Y.Doc body was empty at join time
  isWritable: boolean;  // false for read-only peers
  peerCount: number;    // other peers currently in the room
}

The client uses isEmpty to decide whether to seed initialDocument. Only one peer per fresh room sees isEmpty: true — subsequent joiners adopt the existing state instead. Without this signal, two clients connecting near-simultaneously to a fresh room would both seed and create divergent docs.

Custom client wire format:

[varuint message-type] [type-specific payload]

type 0 (SYNC)       — y-protocols/sync
type 1 (AWARENESS)  — y-protocols/awareness encoded update
type 2 (SESSION)    — varString JSON-encoded SessionPayload
type 3 (ASSETS)     — reserved for content-hashed parts (not yet implemented)

A complete custom client implementation lives in src/server.test.ts (under connectAndSync) and src/permissions.test.ts (under wrapConnection). Production code using the standard y-websocket provider can extend its messageHandlers array to handle type 2:

import { WebsocketProvider, messageHandlers } from "y-websocket";
messageHandlers[2] = (encoder, decoder, provider) => {
  const json = decoding.readVarString(decoder);
  const session = JSON.parse(json);
  // Decide whether to seed initialDocument based on session.isEmpty
};

Room routing

By default the room id is the URL pathname (with leading / stripped). Override resolveRoomId for sub-app routing or auth-derived keys:

new SobreeCollabServer({
  port: 1234,
  resolveRoomId: (req) => {
    const u = new URL(req.url ?? "", "http://x");
    return `${u.pathname.split("/")[1]}/${u.searchParams.get("doc")}`;
  },
});

Lifecycle

  • Rooms are created on first peer join (with state hydrated from persistence if available).
  • Y updates from peers are applied to the room's Y.Doc and broadcast to other peers in the room.
  • Awareness updates flow through unchanged (cursors, presence).
  • Persistence writes are debounced (default 2s) to amortize disk I/O.
  • When the last peer leaves, the room is kept alive for a grace period (default 30s) for reconnects. After that, final state is persisted and the room is destroyed.
  • Graceful server.close() flushes every active room before shutdown.

What's NOT in here

  • Authentication. Provide your own via onConnection.
  • Authorization beyond room-level read/write. Field-level ACL (e.g. "this user can only edit comments") would require a validating relay that parses Y updates. Achievable as a separate middleware that intercepts sync-update messages before they reach the room.
  • Asset offloading. Content-hashed binary parts (images, fonts) riding a side-channel blob store rather than the Y.Doc is not implemented in the server. Today binary parts ride along inside Y.Doc updates. (@sobree/core ships a content-hashed BlobStore, but the server has no asset wire channel — message type 3 is reserved and unhandled.)
  • MCP integration. A Y peer that exposes an LLM-friendly RPC face lives in the separate @sobree/mcp package, not in the relay.

License

MIT.