perplexity-mcp-server/dist/mcp-server/tools/perplexitySearch/registration.js

A Perplexity API Model Context Protocol (MCP) server that unlocks Perplexity's search-augmented AI capabilities for LLM agents. Features robust error handling, secure input validation, and transparent reasoning with the showThinking parameter. Built with

import { BaseErrorCode, McpError } from '../../../types-global/errors.js'; // Added McpError
import { ErrorHandler } from '../../../utils/errorHandler.js';
import { logger } from '../../../utils/logger.js';
import { requestContextService } from '../../../utils/requestContext.js';
import { executePerplexitySearch, PerplexitySearchInputSchema } from './logic.js';
// Removed zod-to-json-schema import
/**
 * Registers the perplexity_search tool with the MCP server.
 * This tool utilizes the Perplexity API to perform a search-augmented query. It takes a natural language query, performs a web search using Perplexity's backend, and then uses an LLM (configured via environment variable) to synthesize an answer based on the search results. Optional parameters allow filtering the web search by recency or domain before the LLM processes the information.
 * @param server - The McpServer instance.
 */
export function registerPerplexitySearchTool(server) {
    const operation = 'registerPerplexitySearchTool';
    const toolName = 'perplexity_search';
    // Create a base context for registration logging
    const registrationContext = requestContextService.createRequestContext({ operation, toolName });
    ErrorHandler.tryCatch(// Use sync version for registration
    () => {
        // Register the tool using server.tool()
        server.tool(toolName, 
        // Add a clear description of what the tool does
        "Performs a search-augmented query using the Perplexity API. Takes a natural language query, performs a web search using Perplexity's backend, and uses an LLM to synthesize an answer based on the search results. Optional parameters include filtering by recency or domain, requesting related questions, and showing the model's thinking process.", 
        // Pass the raw shape of the Zod schema
        PerplexitySearchInputSchema.shape, 
        // --- Tool Handler ---
        // Params are automatically validated by the SDK against the shape
        async (params) => {
            // Create context for this specific tool invocation
            const handlerContext = requestContextService.createRequestContext({
                parentContext: registrationContext, // Link to registration context
                operation: 'HandlePerplexitySearchRequest',
                toolName: toolName,
                params: params // Log validated params
            });
            logger.debug("Handling perplexity search request", handlerContext);
            // Wrap the handler logic in tryCatch for robust error handling
            // No need for separate validation here, SDK handles it based on the shape
            return await ErrorHandler.tryCatch(async () => {
                // Delegate the core processing logic, passing the context
                const response = await executePerplexitySearch(params, handlerContext);
                logger.debug("Perplexity search tool processed successfully", handlerContext);
                // Return the response directly as it's already in McpToolResponse format
                return response;
            }, {
                // Configuration for the error handler specific to this tool call
                operation: 'processing perplexity search handler',
                context: handlerContext, // Pass handler-specific context
                input: params, // Log input parameters on error
                // Provide a custom error mapping for more specific error reporting
                errorMapper: (error) => new McpError(error instanceof McpError ? error.code : BaseErrorCode.INTERNAL_ERROR, `Error processing perplexity search tool: ${error instanceof Error ? error.message : 'Unknown error'}`, { ...handlerContext } // Include context in the McpError
                )
            });
        }); // End of server.tool call
        logger.info(`Tool registered successfully: ${toolName}`, registrationContext);
    }, {
        // Configuration for the error handler wrapping the entire registration
        operation: `registering tool ${toolName}`,
        context: registrationContext, // Context for registration-level errors
        errorCode: BaseErrorCode.INTERNAL_ERROR, // Default error code for registration failure
        // Custom error mapping for registration failures
        errorMapper: (error) => new McpError(error instanceof McpError ? error.code : BaseErrorCode.INTERNAL_ERROR, `Failed to register tool '${toolName}': ${error instanceof Error ? error.message : 'Unknown error'}`, { ...registrationContext } // Include context in the McpError
        ),
        critical: true // Mark registration failure as critical
    }); // End of ErrorHandler.tryCatch for registration
}
;