What Roots Tell a Server

Your server runs as its own process, so it cannot just see the user's project. It needs the client to tell it which folders are fair game. That hint is called a root. 🧭

A root is a single location a client says your server may work within, usually one project folder. It marks the edge of where you should reach.

Most roots are file:// URIs pointing at a directory on disk. The scheme makes it explicit that you are looking at a local folder, not a web address.

Each root can carry an optional name, a short label like "My App". It is just for humans, so logs and prompts read clearly instead of showing raw paths.

A client can expose several roots at once, like a code folder plus a docs folder. Your server should treat the list as the full set of places it may touch.

Roots are guidance, not hard enforcement by the protocol. They tell you where to focus, but it is on you to actually stay inside them. ✋

Roots let a user say "work here, nowhere else." That scoping stops a server from wandering into private files far outside the project the user actually opened.

The client decides what counts as a root, often from the workspace folder the user opened. Your server never sets its own roots; it only reads what it is given.

Roots are not frozen. If the user opens a new folder, the set changes, and a capable client can tell your server the list has been updated.

Roots are an optional client feature. Some hosts offer them, some do not, so a well-built server works gracefully whether the list is full or empty.

On the wire a root entry is tiny: just a uri and an optional name. That small shape is all your server needs to know where it can operate.

Make sure the purpose of a root is clear.

You learned a root is a client-given location, often a file:// folder with an optional name, that scopes where your server should work. The client owns the list, and it can change. 🎯