Skip to content

MCP server for semantic editing of Markdown documents with structural navigation, content manipulation, and YAML frontmatter support.

License

Notifications You must be signed in to change notification settings

KazKozDev/markdown-editor-mcp-server

Repository files navigation

Markdown Editor MCP Server

Tests PyPI version Python 3.10+ License: MIT MCP 2025

MCP server providing tools for structured, semantic editing of Markdown files. Unlike standard text editors, this server understands the logical structure of your documents.

Fully compliant with MCP 2025 Standard - includes Tool Search, Examples, Output Schemas, and Dynamic Capabilities.

Installation

From PyPI (recommended)

pip install markdown-editor-mcp-server

From source

git clone https://github.com/KazKozDev/markdown-editor-mcp-server.git
cd markdown-editor-mcp-server
pip install -e .

Claude Desktop Configuration

Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):

{
  "mcpServers": {
    "markdown-editor": {
      "command": "markdown-editor-mcp-server"
    }
  }
}

Available Tools

Discovery & Scaling

search_tools - Find the right tool

Why LLMs need this: With 15 tools available, finding the right one can be challenging. This meta-tool helps discover which tool to use for a specific task.

What it does: Searches through all available tools and returns the most relevant ones based on your query.

Parameters:

  • query (required): Description of what you want to do

Example:

{"query": "replace text"}

Semantic Editing

get_document_structure - Get document tree

Why LLMs need this: Large Markdown files are hard to navigate. This tool converts the file into a structural tree, allowing the LLM to understand the hierarchy of headings, paragraphs, and lists.

What it does: Parses the Markdown file and returns a JSON tree of all elements with their semantic paths.

Parameters:

  • file_path (required): Path to the .md file
  • depth: Max depth for the tree (default: 2)

Example:

{"file_path": "docs/report.md", "depth": 3}
search_text - Semantic search and navigation

Why LLMs need this: Instead of scrolling through an entire file, the LLM can search for specific keywords and get exact semantic paths (e.g., Project > Deadlines > paragraph 2).

What it does: Searches the document for text and returns paths to the containing elements.

Parameters:

  • file_path (required): Path to the file
  • query (required): Text to search for

Example:

{"file_path": "todo.md", "query": "urgent"}
read_element - Read specific element

Why LLMs need this: Before editing an element, you often need to see its full content. This tool fetches the complete content of a specific block.

What it does: Returns the full content of an element identified by its path.

Parameters:

  • file_path (required): Path to the file
  • path (required): Semantic path to the element

Example:

{"file_path": "notes.md", "path": "Features > list 1"}
replace_content - Precise content replacement

Why LLMs need this: Overwriting entire files is risky and token-expensive. This tool allows the LLM to replace the content of a specific semantic block without affecting the rest of the document.

What it does: Replaces text in a specific element identified by its path.

Parameters:

  • file_path (required): Path to the file
  • path (required): Semantic path (e.g., "Intro > paragraph 1")
  • new_content (required): New text for the block

Example:

{"file_path": "readme.md", "path": "Installation > paragraph 1", "new_content": "Just run pip install."}
insert_element - Add new content

Why LLMs need this: Adding new paragraphs, headings, or lists requires understanding document structure. This tool inserts new elements at the right location.

What it does: Inserts a new block (heading, paragraph, list, code block, or blockquote) before or after an existing element.

Parameters:

  • file_path (required): Path to the file
  • path (required): Reference element path
  • element_type (required): Type (heading, paragraph, list, code_block, blockquote)
  • content (required): Content of the new element
  • where: "before" or "after" (default: "after")
  • heading_level: Level for headings (default: 1)

Example:

{"file_path": "doc.md", "path": "Introduction", "element_type": "paragraph", "content": "New paragraph here."}
delete_element - Remove content

Why LLMs need this: Removing specific blocks without affecting surrounding content requires precision. This tool deletes elements by their semantic path.

What it does: Removes a block from the document.

Parameters:

  • file_path (required): Path to the file
  • path (required): Semantic path to the element to delete

Example:

{"file_path": "draft.md", "path": "Old Section"}
move_element - Structural refactoring

Why LLMs need this: Reorganizing a document is complex. This tool lets the LLM move entire sections (including all nested sub-elements) to a new location with a single command.

What it does: Moves a block of text from source path to target path.

Parameters:

  • file_path (required): Path to the file
  • source_path (required): Path of element to move
  • target_path (required): Reference path for new location
  • where: "before" or "after" the target (default: "after")

Example:

{"file_path": "draft.md", "source_path": "Contacts", "target_path": "Conclusion"}
get_context - Context-aware editing

Why LLMs need this: When editing a single block, LLMs might lose the narrative flow. This tool provides the target element along with its immediate neighbors (before and after).

What it does: Fetches an element and snippets of surrounding blocks.

Parameters:

  • file_path (required): Path to the file
  • path (required): Path to the target element

Example:

{"file_path": "article.md", "path": "Body > paragraph 5"}
update_metadata - YAML Frontmatter management

Why LLMs need this: Many Markdown tools (Obsidian, Jekyll) use YAML metadata at the top. This tool allows the LLM to manage tags, dates, and properties without messing with the body text.

What it does: Updates or adds YAML Frontmatter to the document.

Parameters:

  • file_path (required): Path to the file
  • metadata (required): Dictionary of metadata items

Example:

{"file_path": "post.md", "metadata": {"status": "published", "tags": ["mcp", "ai"]}}
undo - Revert changes

Why LLMs need this: Mistakes happen. This tool allows reverting recent operations without manual file restoration.

What it does: Reverts the last N operations on the document.

Parameters:

  • file_path (required): Path to the file
  • count: Number of operations to undo (default: 1)

Example:

{"file_path": "document.md", "count": 2}

File Operations

list_directory - Explore workspace

What it does: Lists files and folders in a directory, helping the LLM explore the file structure.

Parameters:

  • path: Directory path (default: ".")
create_file - Create new document

What it does: Creates a new file with optional initial content.

Parameters:

  • path (required): Path to the new file
  • content: Initial content (default: "")

Example:

{"path": "./notes/todo.md", "content": "# TODO\n\n- Task 1"}
create_directory - Create folder

What it does: Creates a new directory.

Parameters:

  • path (required): Path to the new directory

Example:

{"path": "./docs/archive"}
delete_item - Cleanup

What it does: Deletes a file or directory.

Development

# Install for development
pip install -e .

# Run tests
pytest tests/

MCP 2025 Standard Features

This server implements all 2025 MCP improvements:

  • Tool Search Tool - Efficiently find the right tool among many options
  • Tool Use Examples - Input parameter examples help LLMs use tools correctly
  • Output Schemas - Structured output definitions for better client integration
  • Dynamic Capabilities - Support for tools/list_changed notifications

If you like this project, please give it a star ⭐

Artem KK | MIT LICENSE