pdfdancer-mcp

Instructions

Implementation Reference

• src/index.ts:201-212 (handler)

  The handler function for the 'help' tool. It renders the help document using renderHelpDocument and returns markdown text content along with the structured HelpDocument data.

  async () => {
      const text = renderHelpDocument(helpDocument);
      return {
          content: [
              {
                  type: 'text' as const,
                  text
              }
          ],
          structuredContent: helpDocument
      };
  }

• src/index.ts:195-213 (registration)

  Registration of the 'help' tool with the MCP server, including tool name, metadata (title and description), and the inline handler function.

  server.registerTool(
      'help',
      {
          title: 'PDFDancer MCP help',
          description: 'Explains the purpose of this server and how to use it.'
      },
      async () => {
          const text = renderHelpDocument(helpDocument);
          return {
              content: [
                  {
                      type: 'text' as const,
                      text
                  }
              ],
              structuredContent: helpDocument
          };
      }
  );

• src/index.ts:58-65 (schema)

  TypeScript interface defining the structure of the HelpDocument used for structured content in the 'help' tool response.

  interface HelpDocument extends JsonObject {
      title: string;
      overview: string;
      sections: Array<{
          heading: string;
          items: string[];
      }>;
  }

• src/index.ts:182-187 (helper)

  Helper function that converts a HelpDocument object into a formatted Markdown string for the 'help' tool's text content.

  function renderHelpDocument(doc: HelpDocument): string {
      const sections = doc.sections
          .map(section => `### ${section.heading}\n${section.items.join('\n')}`)
          .join('\n\n');
      return `# ${doc.title}\n\n${doc.overview}\n\n${sections}`.trim();
  }

• src/index.ts:131-180 (helper)

  Static HelpDocument instance containing the overview, sections, and tool descriptions used by the 'help' tool handler.

  const helpDocument: HelpDocument = {
      title: 'PDFDancer SDK Documentation',
      overview:
          'This MCP server provides coding agents with searchable access to official PDFDancer SDK documentation. Use these tools to discover how to build PDF manipulation applications with pixel-perfect control over PDF documents using PDFDancer SDKs for Python, TypeScript, and Java.',
      sections: [
          {
              heading: 'What is PDFDancer?',
              items: [
                  '- **Pixel-perfect PDF control**: Edit any PDF with exact coordinate positioning and surgical text replacement',
                  '- **Edit existing PDFs**: Modify PDFs you did not create - invoices, forms, contracts, reports from any source',
                  '- **Smart text handling**: Paragraph-aware editing that preserves layout and formatting',
                  '- **Embedded font support**: Add text with embedded fonts for consistent rendering across all viewers',
                  '- **Form manipulation**: Programmatically fill, update, or delete AcroForm fields',
                  '- **Multi-language SDKs**: Available for Python 3.10+, TypeScript (Node.js 20+), and Java 11+',
                  '- **Vector graphics control**: Full control over lines, curves, shapes, and complex drawings'
              ]
          },
          {
              heading: 'Available MCP Tools',
              items: [
                  '- **help**: Display this overview',
                  '- **version**: Get the current version of the pdfdancer-mcp server',
                  '- **search-docs**: Search PDFDancer documentation by keyword to find relevant APIs and examples (max 10 results)',
                  '- **get-docs**: Retrieve complete documentation for a specific route with code examples'
              ]
          },
          {
              heading: 'Search Syntax (Lunr.js)',
              items: [
                  '- **Simple keywords** (OR logic): `Java images add` → matches documents with ANY word',
                  '- **Require ALL words**: `+Java +images +add` → matches only documents with ALL words (recommended for precise results)',
                  '- **Exclude words**: `+React -deprecated` → must have "React", must NOT have "deprecated"',
                  '- **Boost importance**: `authentication^10 tutorial` → prioritizes "authentication" results',
                  '- **Field search**: `title:installation` → searches only in title field',
                  '- **Wildcard**: `install*` → matches "install", "installation", "installer"',
                  '- **Fuzzy search**: `javascript~2` → tolerates up to 2 character differences (typos)',
                  '- **Exact phrase**: `title:"Java images"` → searches for exact phrase in title',
                  '- **Available fields**: title, content, tags, sidebarParentCategories'
              ]
          },
          {
              heading: 'Typical workflow for building PDFDancer applications',
              items: [
                  '1. **Search documentation**: Use `search-docs` with keywords like "authentication", "edit text", "add paragraph", or "forms" to find relevant APIs',
                  '2. **Get detailed docs**: Use `get-docs` with the route from search results to retrieve complete documentation with code examples',
                  '3. **Implement features**: Use the SDK to open existing PDFs, locate elements by text or coordinates, edit content, add new elements, and save results'
              ]
          }
      ],
  };