Open Search MCP

Instructions

Implementation Reference

• src/tools/pdf/research.ts:173-242 (handler)

  Main handler function for pdf_discovery tool. Performs PDF search using PDFProcessor.searchPDFs and returns list of discovered documents with metadata.

  async function pdfDiscovery(args: ToolInput): Promise<ToolOutput> {
    const { 
      query, 
      maxResults = 20, 
      sources = ['all'],
      documentType = 'any'
    } = args;
    
    try {
      logger.info(`Starting PDF discovery for: ${query}`);
      
      if (!query || typeof query !== 'string') {
        throw new Error('Query parameter is required and must be a string');
      }
      
      const pdfProcessor = new PDFProcessor();
      
      const searchOptions: PDFSearchOptions = {
        query,
        maxDocuments: maxResults,
        documentType: documentType as any,
        includeOCR: false, // Discovery doesn't need OCR
        sources: Array.isArray(sources) ? sources : [sources]
      };
      
      const pdfDocuments = await pdfProcessor.searchPDFs(searchOptions);
      
      const result: ToolOutput = {
        success: true,
        data: {
          query,
          documents: pdfDocuments.map(doc => ({
            id: doc.id,
            title: doc.title,
            url: doc.url,
            source: doc.source,
            relevanceScore: doc.relevanceScore,
            downloadUrl: doc.downloadUrl,
            fileSize: doc.fileSize
          })),
          totalFound: pdfDocuments.length,
          searchOptions: {
            documentType,
            sources: searchOptions.sources
          },
          searchedAt: new Date().toISOString()
        },
        metadata: {
          sources: ['pdf-discovery'],
          cached: false
        }
      };
  
      logger.info(`PDF discovery completed: ${pdfDocuments.length} documents found for ${query}`);
      return result;
  
    } catch (error) {
      logger.error(`Failed PDF discovery for ${query}:`, error);
      
      return {
        success: false,
        error: `Failed to discover PDFs: ${error instanceof Error ? error.message : 'Unknown error'}`,
        data: null,
        metadata: {
          sources: ['pdf-discovery'],
          cached: false
        }
      };
    }
  }

• src/tools/pdf/research.ts:406-445 (registration)

  Tool registration: creates the pdf_discovery tool using createTool, assigns inputSchema, and registers it with the registry.

  const pdfDiscoveryTool = createTool(
    'pdf_discovery',
    'Discover PDF documents without full processing - fast PDF search and listing',
    'pdf',
    'pdf-discovery',
    pdfDiscovery,
    {
      cacheTTL: 1800, // 30 minutes cache
      rateLimit: 15,  // 15 requests per minute
      requiredParams: ['query'],
      optionalParams: ['maxResults', 'sources', 'documentType']
    }
  );
  
  pdfDiscoveryTool.inputSchema = {
    type: 'object',
    properties: {
      query: {
        type: 'string',
        description: 'Search query for PDF document discovery'
      },
      maxResults: {
        type: 'number',
        description: 'Maximum number of results to return (default: 20)'
      },
      sources: {
        type: 'array',
        items: { type: 'string' },
        description: 'Sources to search (arxiv, pubmed, web, all)'
      },
      documentType: {
        type: 'string',
        description: 'Type of documents to search for',
        enum: ['academic', 'report', 'manual', 'any']
      }
    },
    required: ['query']
  };
  
  registry.registerTool(pdfDiscoveryTool);

• src/tools/pdf/research.ts:420-443 (schema)

  Input schema definition for the pdf_discovery tool, specifying properties like query, maxResults, sources, documentType.

  pdfDiscoveryTool.inputSchema = {
    type: 'object',
    properties: {
      query: {
        type: 'string',
        description: 'Search query for PDF document discovery'
      },
      maxResults: {
        type: 'number',
        description: 'Maximum number of results to return (default: 20)'
      },
      sources: {
        type: 'array',
        items: { type: 'string' },
        description: 'Sources to search (arxiv, pubmed, web, all)'
      },
      documentType: {
        type: 'string',
        description: 'Type of documents to search for',
        enum: ['academic', 'report', 'manual', 'any']
      }
    },
    required: ['query']
  };