The WorkFlowy MCP Server enables LLM applications to manage, navigate, and transform WorkFlowy outlines through comprehensive tools for node operations, advanced bulk transformations, and hierarchical exploration.
Core Node Operations: Create, read, update, delete, move, complete/uncomplete, and export nodes with support for names, notes, layout modes (bullets/todo/h1/h2/h3), and hierarchical positioning.
NEXUS System - Advanced Bulk Operations:
GLIMPSE - Instant WebSocket-based capture of expanded WorkFlowy views creating terrain and phantom gem files
Corinthian Workflow - Multi-stage process: coarse terrain mapping (scry), selective deepening (ignite shards), revelation (anchor gems/jewels), semantic transformations (JEWELSTORM operations: MOVE/DELETE/RENAME/CREATE, SET_NOTE/ATTRS, SEARCH_REPLACE with regex), and weaving changes back to WorkFlowy
Guided Exploration - Start, step through, resume, and finalize exploration sessions with DFS traversal modes and node labeling
Loading & Export: Load entire trees via WebSocket (glimpse), API (scry), or hybrid methods with depth/size controls; export nodes with all children; convert between JSON and Markdown formats.
ETCH Commands: Create multiple nodes from JSON structures synchronously or asynchronously for large operations.
Job Management: Track status and cancel long-running async operations (ETCH, NEXUS WEAVE).
Backup/Restore: Create, list, restore, and purge Keystone backups of node trees.
Key Limitation: Must navigate hierarchically from root—the WorkFlowy API doesn't support direct search by name/content or jumping to deeply nested nodes. Web interface node IDs are incompatible with API IDs.
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@WorkFlowy MCP Servercreate a new task called 'Finish quarterly report' under my 'Work Projects' node"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
WorkFlowy MCP Server
A Model Context Protocol (MCP) server that integrates WorkFlowy's outline and task management capabilities with LLM applications.
MCP Tools Available
Tool | Description |
| Create new nodes with name, notes, and layout mode |
| Update existing node properties |
| Retrieve a specific node by ID |
| List child nodes of a specific parent |
| Delete a node and its children |
| Mark a node as completed |
| Mark a node as uncompleted |
⚠️ Important Limitations
The WorkFlowy API has significant discovery limitations:
✅ CAN list root-level nodes (call
list_nodeswithout parent_id)✅ CAN navigate down the tree by listing children of discovered nodes
❌ CANNOT search for nodes by name or content
❌ CANNOT jump directly to deeply nested nodes
❌ CANNOT use node IDs from WorkFlowy web URLs (they use different IDs)
Practical Impact:
You must navigate hierarchically from root to find existing nodes
No text search means manually traversing the tree to find specific content
Deep nodes require multiple list operations to reach
The web interface IDs (
workflowy.com/#/abc123) are NOT compatible with API IDs
Quick Start
Prerequisites
Python 3.10 or higher
WorkFlowy account with API access
Claude Desktop or other (local, since it's a python package) MCP-compatible client
Installation
Option 1: Install from PyPI (Recommended)
# Install the package
pip install workflowy-mcpOption 2: Quick Setup Script
# Download and run the setup script
curl -sSL https://raw.githubusercontent.com/yourusername/workflowy-mcp/main/install.sh | bash
# Or on Windows:
# irm https://raw.githubusercontent.com/yourusername/workflowy-mcp/main/install.ps1 | iexOption 3: Manual Installation from Source
# Clone the repository (if you want to contribute or modify)
git clone https://github.com/vladzima/workflowy-mcp.git
cd workflowy-mcp
pip install -e .Configuration
Get your WorkFlowy API key:
From WorkFlowy
Configure client: Edit your client configuration (Claude Desktop example):
Mac:
~/Library/Application Support/Claude/claude_desktop_config.jsonWindows:
%APPDATA%\Claude\claude_desktop_config.json
Add to the
mcpServerssection:{ "mcpServers": { "workflowy": { "command": "python3", "args": ["-m", "workflowy_mcp"], "env": { "WORKFLOWY_API_KEY": "your_actual_api_key_here", // Optional settings (uncomment to override defaults): // "WORKFLOWY_API_URL": "https://workflowy.com/api/v1", // "WORKFLOWY_REQUEST_TIMEOUT": "30", // "WORKFLOWY_MAX_RETRIES": "3", // "WORKFLOWY_RATE_LIMIT_REQUESTS": "60", // "WORKFLOWY_RATE_LIMIT_WINDOW": "60" } } } }Restart your client to load the MCP server
Usage
Once configured, you can use WorkFlowy tools with your agent:
Working with New Nodes
"Create a new WorkFlowy node called 'Project Tasks'"
# Returns: Created node with ID: abc-123-def
"Create a todo item 'Review PR' under parent node abc-123-def"
"Mark the node abc-123-def as completed"
"List all children of node abc-123-def"Navigating Existing Nodes
Since there's no search, you must navigate from root:
"List my root-level WorkFlowy nodes"
# Returns: List of top-level nodes with their IDs
"List children of node abc-123-def"
# Navigate deeper into your outline
"Get details for node abc-123-def"
"Update node abc-123-def with new notes"Note: The node IDs from the web interface URLs are NOT compatible with the API.
Development
Setup Development Environment
# Create virtual environment
python -m venv venv
source venv/bin/activate # On Windows: venv\Scripts\activate
# Install in development mode
pip install -e ".[dev]"
# Run tests
pytest
# Run with coverage
pytest --cov=workflowy_mcp
# Run linting
ruff check src/
mypy src/
black src/ --checkProject Structure
workflowy-mcp/
├── src/
│ └── workflowy_mcp/
│ ├── __init__.py
│ ├── __main__.py # Entry point
│ ├── server.py # FastMCP server & tools
│ ├── config.py # Configuration
│ ├── transport.py # STDIO transport
│ ├── client/
│ │ ├── api_client.py # WorkFlowy API client
│ │ ├── rate_limit.py # Rate limiting
│ │ └── retry.py # Retry logic
│ ├── models/
│ │ ├── node.py # Node models
│ │ ├── requests.py # Request models
│ │ ├── config.py # Config models
│ │ └── errors.py # Error models
│ └── middleware/
│ ├── errors.py # Error handling
│ └── logging.py # Request logging
├── tests/
│ ├── contract/ # Contract tests
│ ├── integration/ # Integration tests
│ ├── unit/ # Unit tests
│ └── performance/ # Performance tests
├── pyproject.toml # Project configuration
├── README.md # This file
├── CONTRIBUTING.md # Contribution guide
├── install.sh # Unix/Mac installer
└── install.ps1 # Windows installerRunning Tests
# Run all tests
pytest
# Run specific test categories
pytest tests/unit/
pytest tests/contract/
pytest tests/integration/
pytest tests/performance/
# Run with coverage report
pytest --cov=workflowy_mcp --cov-report=html
# Run with verbose output
pytest -xvsAPI Reference
Node Structure
{
"id": "unique-node-id",
"name": "Node name", # Text content
"note": "Node notes/description", # Optional notes
"layoutMode": "bullets", # Display mode: bullets, todo, h1, h2, h3
"completedAt": null, # Completion timestamp (null if not completed)
"children": [], # Child nodes array
"createdAt": 1234567890, # Unix timestamp
"modifiedAt": 1234567890 # Unix timestamp
}Error Handling
All tools return a consistent error format:
{
"success": false,
"error": "error_type",
"message": "Human-readable error message",
"context": {...} // Additional error context
}Performance
Automatic rate limiting prevents API throttling
Token bucket algorithm for smooth request distribution
Adaptive rate limiting based on API responses
Connection pooling for efficient HTTP requests
Contributing
See CONTRIBUTING.md for development setup and contribution guidelines.
License
MIT License - see LICENSE file for details.
Support
Acknowledgments
Built with FastMCP framework
Integrates with WorkFlowy API
Compatible with Claude Desktop and other MCP clients