Document Organizer MCP Server

# Document Organizer MCP - Current Status **Last Updated:** 2025-09-30 **Version:** 1.0.0 **Status:** Production (Pending Major Refactor) ## Current Reality ✅ ### What's Actually Working - ✅ PDF → Markdown conversion (Marker engine) - ✅ PDF → Markdown conversion (pymupdf4llm engine) - ✅ Automatic marker output cleaning (table-aware) - ✅ Dependency checking (pymupdf4llm, marker) - ✅ Universal Project Documentation Standard tools (10 tools) - ✅ Document discovery and organization workflow (6 tools) ### Current Tool Inventory (17 Total) **Conversion Tools (2):** - `convert_pdf` - PDF to Markdown with dual engine support - `check_dependency` - Verify/install pymupdf4llm **Document Organization Tools (10 - DEPRECATED):** - `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` ### Known Issues ❌ 1. **Timeout Issue** - `check_dependency` can hit 30s MCP timeout - Root cause: Synchronous Python process spawn without explicit timeout - Impact: Blocks MCP requests during dependency checks - Workaround: None currently - Fix planned: See FEATURE_BACKLOG.md #6 2. **No Markdown → PDF Conversion** - Currently one-way conversion only 3. **Limited Format Support** - Only PDF→Markdown implemented ## Upcoming Major Refactor 🚧 ### Strategic Direction Change **FROM:** Document organization + conversion + project documentation **TO:** Best-in-class file conversion **ONLY** ### Rename Plan - **Old Name:** `document-organizer-mcp` - **New Name:** `file-converter-mcp` - **Rationale:** Focus on core competency (conversion), remove scope creep ### Scope Reduction **Removing:** - All document organization tools (10 tools) - Universal Project Documentation Standard tools (moved to separate MCP) **Keeping & Enhancing:** - Core conversion tools (2 tools → expanding to 6+) - Adding Pandoc for bidirectional conversion - Adding LibreOffice for Office document support **See:** [FEATURE_BACKLOG.md](./FEATURE_BACKLOG.md) for complete refactor plan ## Component Status Matrix | Component | Implementation | Testing | Documentation | Status | |-----------|---------------|---------|---------------|--------| | PDF→MD (Marker) | ✅ Complete | 🟡 Basic | ✅ Complete | 100% | | PDF→MD (pymupdf4llm) | ✅ Complete | 🟡 Basic | ✅ Complete | 100% | | Dependency Checks | ⚠️ Timeout Issues | ❌ None | ✅ Complete | 70% | | Document Discovery | ✅ Complete | ❌ None | ✅ Complete | 100% | | Document Organization | ✅ Complete | ❌ None | ✅ Complete | 100% | | Project Doc Standard | ✅ Complete | ❌ None | ✅ Complete | 100% | | **Pandoc Integration** | ❌ Backlog | ❌ None | 📋 Planned | 0% | | **Format Expansion** | ❌ Backlog | ❌ None | 📋 Planned | 0% | ## Recent Key Decisions - **2025-09-30:** Audit revealed timeout issues in dependency checking - **2025-09-30:** Decision to refactor into conversion-focused MCP - **2025-09-30:** Created feature backlog for refactor roadmap - **2025-09-30:** Identified Pandoc + Marker as optimal conversion stack ## Next Priority Actions ### Immediate (This Week) 1. Review and approve FEATURE_BACKLOG.md refactor plan 2. Decide on migration timeline for organization tools 3. Begin Pandoc integration research/prototyping ### Short-term (Next 2 Weeks) 1. Implement timeout fixes for dependency checking 2. Add Pandoc integration for Markdown→PDF 3. Create conversion quality benchmarking suite 4. Plan rename from `document-organizer` → `file-converter` ### Medium-term (Next Month) 1. Add LibreOffice integration for Office documents 2. Implement conversion engine selection logic 3. Remove deprecated organization tools 4. Update all documentation for new scope ## Development Environment **Dependencies:** - Node.js 18+ - TypeScript 5.0+ - Python 3 (for marker, pymupdf4llm) - marker-pdf (`pip install marker-pdf`) - pymupdf4llm (`pip install pymupdf4llm`) **Planned Additions:** - Pandoc (`apt install pandoc`) - WeasyPrint (`pip install weasyprint`) - LibreOffice (`apt install libreoffice`) **Build & Test:** ```bash npm install npm run build npm start # Test MCP directly echo '{"jsonrpc":"2.0","id":1,"method":"tools/list"}' | node dist/index.js ``` ## Integration Status **Currently Integrated With:** - metaMCP-RAG server (RAG-based tool discovery) - Claude Code (via MCP protocol) **Future Integrations:** - Will need config updates after rename to `file-converter` - Consider standalone CLI version for batch operations ## Performance Metrics **Current Performance:** - Marker conversion: ~2-5 seconds per page - pymupdf4llm: ~1-2 seconds per page - Dependency check: 2-30 seconds (timeout issue) **Target Performance (Post-Refactor):** - Pandoc MD→PDF: <1 second per page - Batch operations: Parallel processing support - Dependency checks: <100ms (cached results) ## Documentation Status - ✅ README.md - Comprehensive user guide - ✅ CURRENT_STATUS.md - This file - ✅ FEATURE_BACKLOG.md - Refactor roadmap - ❌ ACTIVE_PLAN.md - Needed for refactor execution - ❌ .claude-instructions.md - Project-specific guidelines - ❌ API documentation - Tool schemas only ## Blockers & Dependencies **No Current Blockers** **External Dependencies:** - Marker project maintenance (GitHub: datalab-to/marker) - Pandoc development (stable, mature project) - LibreOffice headless mode stability --- **Remember:** This MCP is undergoing strategic refocusing. Current functionality remains stable, but significant changes are planned. See FEATURE_BACKLOG.md for details.