document-mcp 0.0.1

A MCP server for managing structured Markdown documents

Document MCP Server

A Model Context Protocol (MCP) server for managing structured Markdown documents. This server provides tools to create, read, update, and analyze documents composed of multiple chapters.

Installation

pip install document-mcp

Quick Start

# Start the MCP server
python -m document_mcp.doc_tool_server sse --host localhost --port 3001

Overview

This MCP server treats "documents" as directories containing multiple "chapter" files (Markdown .md files). Chapters are ordered alphanumerically by their filenames (e.g., 01-introduction.md, 02-main_content.md).

Document Structure

documents_storage/           # Root directory for all documents
├── my_book/                     # A document (directory)
│   ├── 01-introduction.md       # Chapter 1 (alphanumeric ordering)
│   ├── 02-main_content.md       # Chapter 2
│   ├── 03-conclusion.md         # Chapter 3
│   └── _manifest.json           # Optional: For future explicit chapter ordering
└── research_paper/              # Another document
    ├── 00-abstract.md
    ├── 01-methodology.md
    └── 02-results.md

Configuration

The server uses the following environment variables:

• DOCUMENT_ROOT_DIR: Root directory for storing documents (default: documents_storage/)

Running the Server

The server supports both HTTP SSE and stdio transports. HTTP SSE is the default and recommended transport.

HTTP SSE Transport (Recommended)

# Run with HTTP SSE transport (default)
python -m document_mcp.doc_tool_server sse --host localhost --port 3001

# Or specify arguments explicitly
python -m document_mcp.doc_tool_server sse --host 0.0.0.0 --port 8000

Stdio Transport

# Run with stdio transport
python -m document_mcp.doc_tool_server stdio

MCP Tools Reference

The server exposes the following tools via the Model Context Protocol:

Document Management

Tool	Parameters	Description
list_documents	-	Lists all available documents with metadata
create_document	document_name: str	Creates a new document directory
delete_document	document_name: str	Deletes a document and all its chapters

Chapter Management

Tool	Parameters	Description
list_chapters	document_name: str	Lists all chapters in a document, ordered by filename
create_chapter	document_name: str, chapter_name: str, initial_content: str = ""	Creates a new chapter file
delete_chapter	document_name: str, chapter_name: str	Deletes a chapter from a document

Content Operations

Tool	Parameters	Description
read_chapter_content	document_name: str, chapter_name: str	Reads the content and metadata of a specific chapter
read_paragraph_content	document_name: str, chapter_name: str, paragraph_index_in_chapter: int	Reads a specific paragraph from a chapter
read_full_document	document_name: str	Reads the entire document, concatenating all chapters
write_chapter_content	document_name: str, chapter_name: str, new_content: str	Overwrites the entire content of a chapter
modify_paragraph_content	document_name: str, chapter_name: str, paragraph_index: int, new_paragraph_content: str, mode: str	Modifies a paragraph (replace, insert_before, insert_after, delete)
append_paragraph_to_chapter	document_name: str, chapter_name: str, paragraph_content: str	Appends a new paragraph to the end of a chapter

Text Operations

Tool	Parameters	Description
replace_text_in_chapter	document_name: str, chapter_name: str, text_to_find: str, replacement_text: str	Replaces all occurrences of text in a specific chapter
replace_text_in_document	document_name: str, text_to_find: str, replacement_text: str	Replaces all occurrences of text throughout all chapters

Analysis Tools

Tool	Parameters	Description
get_chapter_statistics	document_name: str, chapter_name: str	Retrieves statistics (word count, paragraph count) for a chapter
get_document_statistics	document_name: str	Retrieves aggregated statistics for an entire document

Search Tools

Tool	Parameters	Description
find_text_in_chapter	document_name: str, chapter_name: str, query: str, case_sensitive: bool = False	Finds paragraphs containing the query string in a specific chapter
find_text_in_document	document_name: str, query: str, case_sensitive: bool = False	Finds paragraphs containing the query string across all chapters

Data Models

The server uses Pydantic models for structured data exchange:

• DocumentInfo: Metadata for a document
• ChapterMetadata: Metadata for a chapter
• ChapterContent: Full content and metadata of a chapter
• ParagraphDetail: Information about a specific paragraph
• FullDocumentContent: Complete content of a document
• StatisticsReport: Word and paragraph count statistics
• OperationStatus: Success/failure status for operations

Requirements

• Python 3.8+
• fastapi
• uvicorn[standard]
• pydantic-ai
• mcp[cli]
• python-dotenv
• google-generativeai

Examples and Documentation

For comprehensive examples, tutorials, and usage guides, visit the GitHub repository.

License

MIT License

Links

• GitHub Repository: https://github.com/document-mcp/document-mcp
• Bug Reports: GitHub Issues