Struct LspServerConfig 

pub struct LspServerConfig {

    pub language_id: LanguageId,
    pub command: LaunchCommand,
    pub args: ServerArgs,
    pub env: ServerEnv,
    pub file_patterns: Vec<FilePattern>,
    pub initialization_options: Option<Value>,
    pub settings: Option<LspSettings>,
    pub timeout_seconds: TimeoutSecs,
    pub request_timeout_seconds: TimeoutSecs,
    pub heuristics: Option<ServerHeuristics>,
    pub name: Option<ServerId>,
    pub handles: Option<ToolSet>,
    pub indexing: IndexingPolicy,
}

Configuration for a single LSP server.

Serialization goes through a raw mirror of this struct, which reports an invalid file_patterns, env or args entry with the id of the server that holds it and builds the LaunchCommand from the command and selection keys.

Fields

language_id: LanguageId

Language identifier (e.g., “rust”, “python”, “typescript”).

command: LaunchCommand

Command to start the LSP server, with whether mcpls may replace it.

args: ServerArgs

Arguments to pass to the LSP server command.

env: ServerEnv

Environment variables for the LSP server process.

file_patterns: Vec<FilePattern>

File patterns this server handles (glob patterns).

initialization_options: Option<Value>

LSP initialization options (server-specific).

For typescript-language-server, None lets mcpls pin the tsserver bundled next to the server (tsserver.path) so a workspace-supplied tsserver is not run. Setting tsserver.path here overrides the pin, for example to opt back in to the workspace’s TypeScript. Setting any other options disables the pin: mcpls does not merge into user options. TypeScript 7 and later ship no tsserver, so there is nothing to pin: with only that installed the server fails to initialize, and the error says so. A native tsc --lsp --stdio command gets no pin and no generated options. See SECURITY.md for the trust model.

settings: Option<LspSettings>

Per-server settings pushed after initialized via workspace/didChangeConfiguration and served on workspace/configuration. Top-level dotted keys are expanded into nested objects; keys inside values are left untouched.

timeout_seconds: TimeoutSecs

Handshake timeout in seconds: bounds the initialize request during server startup. Does not affect individual tool-call requests sent after initialization; see Self::request_timeout_seconds for that. The LSP server’s shutdown request during teardown uses a separate, fixed 5-second timeout that is not configurable by this field.

request_timeout_seconds: TimeoutSecs

Per-request timeout in seconds, applied to each LSP request issued while translating an MCP tool call (hover, definition, references, etc.).

This bounds a single request attempt, not a whole tool call: on a -32801 (ContentModified) or -32802 (ServerCancelled) response, crate::lsp::LspClient::request retries up to 4 attempts with backoff (one shared budget across both codes), so the worst-case latency for one tool call is 4 * request_timeout_seconds + 3.5 seconds. Completion requests are further capped at 10 seconds regardless of this value; see crate::lsp::LspClient::completion_timeout.

heuristics: Option<ServerHeuristics>

Heuristics for determining if this server should be spawned. If not specified, the server will always attempt to spawn.

name: Option<ServerId>

Human-readable server identity used as the routing key.

Defaults to language_id when omitted (see Self::id). Must be unique across all applicable servers in a workspace, regardless of language: this is what lets two servers share one language_id (e.g. pyright and pylsp both for python) without one silently overwriting the other in the maps keyed by ServerId.

handles: Option<ToolSet>

Tools this server handles.

None means this server is a catch-all: it serves every tool not explicitly claimed by another server for the same language. Some(list) restricts the server to exactly those tools.

indexing: IndexingPolicy

Workspace-indexing readiness policy for this server (P4 escape hatch). Defaults to IndexingPolicy::Auto: readiness is tracked normally from whatever signals the server sends (rust-analyzer’s experimental/serverStatus, or a generic $/progress sequence). Set to "disabled" for a server whose signal shape doesn’t fit this tracker’s assumptions – see IndexingPolicy::Disabled.

Implementations

impl LspServerConfig

pub fn id(&self) -> ServerId

The routing identity of this server: name if set, otherwise language_id.

This is the key used across Translator’s client/server maps, so two servers for the same language must set distinct names or they collide (see ToolRouter::from_configs for the enforcement).

pub fn rust_analyzer() -> Self

Create a default configuration for rust-analyzer.

pub fn pyright() -> Self

Create a default configuration for pyright.

pub fn typescript() -> Self

Create a default configuration for TypeScript language server.

initialization_options stays None here; at startup mcpls fills in the tsserver pin when it can resolve the server’s bundled tsserver (see Self::initialization_options). The entry is ServerSelection::Auto, so mcpls may start the native TypeScript server instead when TypeScript 7 is installed outside the workspace.

pub fn gopls() -> Self

Create a default configuration for gopls.

pub fn clangd() -> Self

Create a default configuration for clangd.

pub fn zls() -> Self

Create a default configuration for zls.

Trait Implementations

impl Clone for LspServerConfig

fn clone(&self) -> Self

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for LspServerConfig

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for LspServerConfig

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for LspServerConfig

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.