Struct ServerRuntime 

pub struct ServerRuntime<S, Tr>
where
    Tr: Transport,
{ /* private fields */ }

Server runtime that handles the message loop.

This runtime manages the connection lifecycle, routes requests to handlers, and coordinates response delivery.

Implementations

impl<S, Tr> ServerRuntime<S, Tr>

where S: RequestRouter + Send + Sync, Tr: Transport + 'static, <Tr as Transport>::Error: Into<McpError>,

pub const fn state(&self) -> &Arc<ServerState>

Get the server state.

pub fn notifier(&self) -> ServerNotifier

Get a cloneable ServerNotifier for sending server-initiated notifications (e.g. tools/list_changed) from outside a request context.

Call this before spawning run; the returned handle shares the runtime’s transport and can be used from any task.

pub async fn run(&self) -> Result<(), McpError>

Run the server message loop.

This method runs until the connection is closed or an error occurs.

Requests are processed concurrently (interleaved on this task) up to config.max_concurrent_requests in flight at once; once that limit is reached, no new messages are accepted until an in-flight request completes (backpressure). Each request runs with panic isolation, so a panicking handler returns a JSON-RPC internal error instead of tearing down the connection. Notification hooks run concurrently too, so a hook that issues its own server-to-client request does not deadlock the loop.

impl<H, T, R, P, K, Tr> ServerRuntime<Server<H, T, R, P, K>, Tr>

where H: ServerHandler + Send + Sync, T: Send + Sync, R: Send + Sync, P: Send + Sync, K: Send + Sync, Tr: Transport + 'static, <Tr as Transport>::Error: Into<McpError>,

pub fn new( server: Server<H, T, R, P, K>, transport: Tr, ) -> ServerRuntime<Server<H, T, R, P, K>, Tr>

Create a new server runtime.

pub fn with_config( server: Server<H, T, R, P, K>, transport: Tr, config: RuntimeConfig, ) -> ServerRuntime<Server<H, T, R, P, K>, Tr>

Create a new server runtime with custom configuration.