Document Organizer MCP Server

# Document Organizer → File Converter MCP - Feature Backlog **Last Updated:** 2025-09-30 **Status:** ACTIVE ## Upcoming Major Refactor: Conversion-Focused Redesign ### Strategic Direction This MCP will be refactored to focus **exclusively on high-quality file conversion**, removing document organization features to create a specialized, best-in-class conversion tool. --- ## High Priority Features ### 1. Add Pandoc Integration for Bidirectional Markdown↔PDF **Priority:** HIGH **Status:** Backlog **Rationale:** Keep Marker for PDF→Markdown (excellent quality), add Pandoc for Markdown→PDF conversion **Implementation Requirements:** - Add Pandoc engine alongside existing Marker/pymupdf4llm - Support multiple PDF engines via Pandoc: - WeasyPrint (recommended default for HTML→PDF) - Typst (modern, fast alternative) - LaTeX engines (pdflatex, lualatex, xelatex) for academic quality - Bidirectional conversion workflow: - PDF → Markdown (Marker) - Markdown → PDF (Pandoc with configurable engine) - Configuration options for PDF quality/style **Expected Tools:** - `convert_markdown_to_pdf` - New tool using Pandoc - Enhanced `convert_pdf` with round-trip validation option **Research Notes:** - Pandoc supports 40+ formats (future expansion potential) - WeasyPrint: Python-based, CSS styling support - Typst: Gaining popularity in 2025, faster than LaTeX - LaTeX: Highest quality for academic/professional documents --- ### 2. Expand Document Conversion Capabilities (High-Quality Focus) **Priority:** HIGH **Status:** Backlog **Rationale:** Aggregate best-in-class conversion tools for comprehensive format support **Target Conversion Capabilities:** #### Additional PDF Converters - **Docling** - Flawless table preservation, complex layouts - **MinerU** - Academic papers, research documents - **Mistral Document AI** (optional, API-based) - Highest accuracy, commercial #### Office Document Conversions - **DOC/DOCX ↔ PDF** - LibreOffice headless mode (current best practice) - Command: `libreoffice --headless --convert-to pdf file.docx` - Batch support with wildcards - **XLSX/XLS ↔ CSV/PDF** - LibreOffice for Excel conversions - Pandas for CSV operations (Python) #### Additional Format Support (Future) - **HTML ↔ Markdown** - Via Pandoc - **reStructuredText ↔ Markdown** - Via Pandoc - **LaTeX ↔ PDF** - Via Pandoc + LaTeX engines - **DOCX ↔ Markdown** - Via Pandoc **Implementation Strategy:** 1. **Phase 1:** Pandoc + LibreOffice (covers 80% of use cases) 2. **Phase 2:** Add Docling/MinerU for specialized PDF needs 3. **Phase 3:** Explore API-based premium converters (Mistral, Mathpix) **Quality Benchmarking:** - Implement conversion quality tests - Compare output across engines for same input - Document when to use which engine --- ### 3. Remove Document Organization Tools (Scope Reduction) **Priority:** HIGH **Status:** Backlog **Rationale:** Focus exclusively on conversion, remove organization features **Tools to REMOVE:** - ❌ `document_organizer__discover_pdfs` - ❌ `document_organizer__check_conversions` - ❌ `document_organizer__convert_missing` - ❌ `document_organizer__analyze_content` - ❌ `document_organizer__organize_structure` - ❌ `document_organizer__full_workflow` - ❌ `document_organizer__init_project_docs` - ❌ `document_organizer__archive_plan` - ❌ `document_organizer__validate_doc_structure` - ❌ `document_organizer__create_weekly_handoff` **Tools to KEEP (and enhance):** - ✅ `convert_pdf` - Core PDF→Markdown conversion - ✅ `check_dependency` - Dependency validation - ✅ NEW: `convert_markdown_to_pdf` - Pandoc-based conversion - ✅ NEW: `convert_document` - Universal converter dispatcher - ✅ NEW: `list_supported_formats` - Capability discovery - ✅ NEW: `convert_batch` - Batch conversion operations **Migration Notes:** - Document organization features could be split to separate MCP if needed - Universal Project Documentation Standard tools (`init_project_docs`, etc.) could move to dedicated "project-template" MCP --- ### 4. Rename MCP Server: `document-organizer` → `file-converter` **Priority:** HIGH **Status:** Backlog **Rationale:** Name should reflect focused conversion purpose **Rename Checklist:** - [ ] Package name: `document-organizer-mcp` → `file-converter-mcp` - [ ] Directory: `document-organizer-mcp/` → `file-converter-mcp/` - [ ] Binary name: `document-organizer-mcp` → `file-converter-mcp` - [ ] Server name in code: `'document-organizer'` → `'file-converter'` - [ ] Tool prefixes: `document_organizer__*` → `convert__*` or no prefix - [ ] README.md updates - [ ] GitHub repository rename - [ ] npm package rename (if published) - [ ] MCP configuration updates in: - metaMCP-RAG server config - Claude Code global config - User documentation **Breaking Changes:** - All existing tool names will change - Configuration updates required for all users - Consider deprecation period with dual naming support --- ## Medium Priority Features ### 5. Add Conversion Quality Validation **Priority:** MEDIUM **Status:** Backlog **Features:** - Round-trip conversion testing (PDF→MD→PDF comparison) - Content validation (character count, image preservation) - Table structure verification - Format integrity checks --- ### 6. Enhanced Error Handling & Timeout Management **Priority:** MEDIUM **Status:** Backlog **Related to:** Issue discovered in audit - dependency check timeout **Improvements:** - Add explicit timeout parameters to all spawn() calls - Cache dependency check results (Redis/in-memory) - Make dependency checking lazy (only when needed) - Async checking with proper error handling - Graceful degradation when engines unavailable **Specific Fixes:** - Line 134-172: Add timeout to `checkPymupdf4llm()` - Line 174-212: Add timeout to `checkMarker()` - Cache results for 5-10 minutes to avoid repeated checks --- ## Low Priority / Future Considerations ### 7. Conversion Engine Performance Benchmarking **Priority:** LOW **Status:** Backlog **Metrics to Track:** - Processing time per page - Memory usage - Output file size - Quality scores (manual evaluation) --- ### 8. API-Based Premium Converter Integration **Priority:** LOW **Status:** Research **Services to Evaluate:** - Mistral Document AI (highest accuracy) - Mathpix (excellent STEM/math support) - CloudConvert API (40+ formats) **Considerations:** - Cost per conversion - API rate limits - Privacy/security for sensitive documents - Offline capability requirements --- ## Completed Features _(None yet - this is the initial backlog)_ --- ## Notes & Considerations ### Design Principles 1. **Quality over Speed** - Prioritize conversion accuracy 2. **Engine Selection Guidance** - Help users choose right tool for job 3. **Graceful Fallbacks** - Multiple engines for resilience 4. **Transparent Results** - Report which engine was used, conversion stats ### Technical Debt - Current ES module timeout issues (lines 134-212) - No caching for expensive dependency checks - Limited batch processing capabilities - No conversion quality metrics ### Future Directions - Consider GUI for conversion configuration - Integration with document management systems - OCR capabilities for scanned PDFs - Specialized converters for code documentation (Sphinx, JSDoc, etc.)