WorkFlowy MCP Server
# 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 |
|------|-------------|
| `workflowy_create_node` | Create new nodes with name, notes, and layout mode |
| `workflowy_update_node` | Update existing node properties |
| `workflowy_get_node` | Retrieve a specific node by ID |
| `workflowy_list_nodes` | List child nodes of a specific parent |
| `workflowy_delete_node` | Delete a node and its children |
| `workflowy_complete_node` | Mark a node as completed |
| `workflowy_uncomplete_node` | Mark a node as uncompleted |
## ⚠️ Important Limitations
**The WorkFlowy API has significant discovery limitations:**
- ✅ **CAN** list root-level nodes (call `list_nodes` without 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)
```bash
# Install the package
pip install workflowy-mcp
```
#### Option 2: Quick Setup Script
```bash
# 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 | iex
```
#### Option 3: Manual Installation from Source
```bash
# 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
1. **Get your WorkFlowy API key**:
- From [WorkFlowy](https://beta.workflowy.com/api-key)
1. **Configure client**:
Edit your client configuration (Claude Desktop example):
- Mac: `~/Library/Application Support/Claude/claude_desktop_config.json`
- Windows: `%APPDATA%\Claude\claude_desktop_config.json`
Add to the `mcpServers` section:
```json
{
"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"
}
}
}
}
```
1. **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
```bash
# 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/ --check
```
### Project 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 installer
```
### Running Tests
```bash
# 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 -xvs
```
## API Reference
### Node Structure
```python
{
"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:
```json
{
"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](CONTRIBUTING.md) for development setup and contribution guidelines.
## License
MIT License - see [LICENSE](LICENSE) file for details.
## Support
- [Report bugs](https://github.com/vladzima/workflowy-mcp/issues)
- [Request features](https://github.com/vladzima/workflowy-mcp/issues)
## Acknowledgments
- Built with [FastMCP](https://github.com/jlowin/fastmcp) framework
- Integrates with [WorkFlowy API](https://beta.workflowy.com/api-reference/)
- Compatible with [Claude Desktop](https://claude.ai/desktop) and other MCP clients
TDQS
Scored across 35 tools
There is significant overlap and confusion among tools, especially between deprecated and current versions (e.g., workflowy_create_single_node vs. workflowy_etch), and between similar NEXUS tools (e.g., nexus_glimpse, nexus_scry, workflowy_glimpse, workflowy_scry). The NEXUS tools have cryptic, overlapping purposes (e.g., nexus_anchor_gems, nexus_anchor_jewels) that are hard to distinguish, leading to high misselection risk.
Naming is mixed with some consistency: most WorkFlowy tools follow a workflowy_verb_noun pattern, and NEXUS tools use nexus_verb_noun, but there are deviations like generate_markdown_from_json and deprecated tools with __WARNING__ suffixes. The patterns are readable but not fully uniform across the set.
With 35 tools, the count is excessive for a WorkFlowy server, especially given many deprecated and overlapping tools. This bloats the interface and makes it hard for agents to navigate efficiently, indicating poor scoping and unnecessary complexity.
The tool set covers core WorkFlowy operations (CRUD, move, complete) and advanced NEXUS features (exploration, transformation, weaving), with minor gaps like bulk updates. However, the domain is well-covered despite some redundancy, allowing agents to perform most tasks.