Struct McpClient 

pub struct McpClient { /* private fields */ }

MCP Client wrapper for connecting to MCP servers

Manages a connection to an MCP server and provides methods to:

• Discover available tools
• Execute tools remotely

Example (when implemented)

// Connect to an MCP server via stdio
let client = McpClient::new_stdio("npx", &["-y", "@modelcontextprotocol/server-filesystem", "/tmp"]).await?;

// Discover tools
let tools = client.list_tools().await?;
println!("Found {} tools", tools.len());

Implementations

impl McpClient

pub async fn new_stdio( command: &str, args: &[&str], ) -> Result<Arc<Self>, String>

Create a new MCP client connected to a server via stdio

Arguments

• command - The command to run (e.g., “npx”, “python”, “node”)
• args - Arguments to pass (e.g., [“-y”, “@modelcontextprotocol/server-filesystem”, “/path”])

Example MCP Servers

• Filesystem: npx -y @modelcontextprotocol/server-filesystem /tmp
• SQLite: npx -y @modelcontextprotocol/server-sqlite --db-path ./data.db
• Web: npx -y @modelcontextprotocol/server-fetch

pub async fn list_tools(&self) -> Result<Vec<McpToolInfo>, String>

Discover all tools available on the connected MCP server

Sends a tools/list request to the MCP server and parses the response.

pub async fn call_tool( &self, name: &str, arguments: HashMap<String, JsonValue>, ) -> Result<JsonValue, String>

Call a tool on the MCP server

Sends a tools/call request with the given arguments and waits for the result.