Struct VisibilityLayer 

pub struct VisibilityLayer<H> { /* private fields */ }

Progressive disclosure through component visibility control. A visibility layer that wraps an McpHandler and filters components.

This allows per-session control over which tools, resources, and prompts are visible to clients through the list_* methods.

Warning: Session overrides stored in this layer must be manually cleaned up via clear_session or by using a VisibilitySessionGuard to prevent unbounded memory growth.

Implementations

impl<H: McpHandler> VisibilityLayer<H>

pub fn new(inner: H) -> Self

Create a new visibility layer wrapping the given handler.

pub fn with_disabled(self, filter: ComponentFilter) -> Self

Disable components matching the filter globally.

This affects all sessions unless explicitly enabled per-session.

pub fn disable_tags<I, S>(self, tags: I) -> Self

where I: IntoIterator<Item = S>, S: Into<String>,

Disable components with the given tags globally.

pub fn enable_for_session(&self, session_id: &str, tags: &[String])

Enable components with the given tags for a specific session.

pub fn disable_for_session(&self, session_id: &str, tags: &[String])

Disable components with the given tags for a specific session.

pub fn clear_session(&self, session_id: &str)

Clear all session-specific overrides.

pub fn session_guard( &self, session_id: impl Into<String>, ) -> VisibilitySessionGuard

Create an RAII guard that automatically cleans up session state on drop.

This is the recommended way to manage session visibility lifetime.

Example

async fn handle_connection<H: McpHandler>(layer: &VisibilityLayer<H>, session_id: &str) {
    let _guard = layer.session_guard(session_id);

    layer.enable_for_session(session_id, &["admin".to_string()]);

    // ... handle requests ...

} // State automatically cleaned up here

pub fn active_sessions_count(&self) -> usize

Get the number of active sessions with visibility overrides.

This is useful for monitoring memory usage.

pub fn inner(&self) -> &H

Get a reference to the inner handler.

pub fn inner_mut(&mut self) -> &mut H

Get a mutable reference to the inner handler.

pub fn into_inner(self) -> H

Unwrap the layer and return the inner handler.

Trait Implementations

impl<H: Clone> Clone for VisibilityLayer<H>

fn clone(&self) -> VisibilityLayer<H>

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl<H: Debug> Debug for VisibilityLayer<H>

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl<H: McpHandler> McpHandler for VisibilityLayer<H>

fn server_info(&self) -> ServerInfo

Returns server information (name, version, description, etc.)

fn list_tools(&self) -> Vec<Tool>

Returns all available tools.

fn list_resources(&self) -> Vec<Resource>

Returns all available resources.

fn list_prompts(&self) -> Vec<Prompt>

Returns all available prompts.

fn call_tool<'a>( &'a self, name: &'a str, args: Value, ctx: &'a RequestContext, ) -> impl Future<Output = McpResult<ToolResult>> + MaybeSend + 'a

Calls a tool by name with the given arguments.

fn read_resource<'a>( &'a self, uri: &'a str, ctx: &'a RequestContext, ) -> impl Future<Output = McpResult<ResourceResult>> + MaybeSend + 'a

Reads a resource by URI.

fn get_prompt<'a>( &'a self, name: &'a str, args: Option<Value>, ctx: &'a RequestContext, ) -> impl Future<Output = McpResult<PromptResult>> + MaybeSend + 'a

Gets a prompt by name with optional arguments.

fn list_tasks<'a>( &'a self, _cursor: Option<&'a str>, _limit: Option<usize>, _ctx: &'a RequestContext, ) -> impl Future<Output = Result<ListTasksResult, McpError>> + MaybeSend + 'a

Lists all active and recent tasks.

fn get_task<'a>( &'a self, _task_id: &'a str, _ctx: &'a RequestContext, ) -> impl Future<Output = Result<Task, McpError>> + MaybeSend + 'a

Gets the current state of a specific task.

fn cancel_task<'a>( &'a self, _task_id: &'a str, _ctx: &'a RequestContext, ) -> impl Future<Output = Result<Task, McpError>> + MaybeSend + 'a

Cancels a running task.

fn get_task_result<'a>( &'a self, _task_id: &'a str, _ctx: &'a RequestContext, ) -> impl Future<Output = Result<Value, McpError>> + MaybeSend + 'a

Gets the result of a completed task.

fn on_initialize( &self, ) -> impl Future<Output = Result<(), McpError>> + MaybeSend

Called when the server is initialized.

fn on_shutdown(&self) -> impl Future<Output = Result<(), McpError>> + MaybeSend

Called when the server is shutting down.