Fortitude Framework

Fortitude is a web framework that enables server-side components defined as Pydantic models, which can be used as input and output schemas for endpoints registered as tools in an external registry.

Features

• Define data models with Pydantic
• Automatic CRUD API endpoints
• Server-side NextJS UI components
• Tool registry integration
• CLI for project management
• MCP (Model Context Protocol) integration for LLM sampling
• Advanced Rails-like scaffolding for rapid development
• Domain-driven design support
• Microservices architecture
• Database migrations

Installation

# Clone the repository
git clone https://github.com/your-username/fortitude.git
cd fortitude

# Install dependencies
pip install -e .

Quick Start

# Create a new project
fort new myproject
cd myproject

# Install dependencies
pip install -r requirements.txt
cd ui && npm install && cd ..

# Generate a model
fort model User

# Start the servers
fort start

Project Structure

myproject/
├── ui/                    # NextJS UI server
│   ├── app/               # App router
│   ├── components/        # React components
│   ├── public/            # Static assets
│   └── package.json
├── backend/               # Backend API server
│   ├── models/            # Pydantic models
│   ├── endpoints/         # API endpoints
│   └── mcp/               # MCP integration
├── main.py                # Main application entry point
└── requirements.txt

Creating Models

Models are defined as Pydantic classes that inherit from FortitudeBaseModel:

from fortitude.backend.models import FortitudeBaseModel
from pydantic import Field
from typing import Optional

class User(FortitudeBaseModel):
    name: str
    email: str
    age: Optional[int] = None

Register the model in your main application:

from backend.models.user import User
app.register_model(User)

Advanced Rails-like Scaffolding

Fortitude offers powerful Rails-like scaffolding capabilities with both standard and advanced options:

Basic Resource Generation

# Generate a complete resource scaffold
fort resource Product

This creates:

• Pydantic model
• CRUD controller
• Service layer
• UI components (list, form, detail views)
• MCP integrations
• Tests

Advanced Resource Generation

# Generate an advanced resource scaffold with relationships, validation, and more
fort resource Product --advanced

The advanced scaffold includes:

• Models with relationships, validations, and computed properties
• Full-featured controllers with authentication
• Service layer with caching, transactions, and error handling
• Advanced UI components with filtering, sorting, pagination
• Comprehensive test suite
• Deployment configurations

Domain-Driven Design

# Generate a domain-driven design scaffold
fort domain Store User:name,email,age Product:name,price,stock Order

This creates a complete domain structure with:

• Core domain models
• Value objects
• Repositories
• Domain services
• Application services
• Aggregates
• Entity relationships

Microservices

# Generate microservices
fort microservice auth --type api
fort microservice worker --type worker
fort microservice gateway --type gateway
fort microservice ai-assistant --type mcp-server

Generates specialized microservices with:

• Containerization
• API gateways
• Service discovery
• Health checks
• Message queues

Custom Scaffolds

For maximum flexibility, generate and customize your own scaffolds:

# Generate a scaffold configuration (standard or advanced)
fort generate Task --advanced --output task_scaffold.py

# Edit the configuration file to customize everything
# Then create the scaffold
fort scaffold task_scaffold.py --advanced

MCP Integration

Fortitude provides comprehensive MCP support with both client and server capabilities:

Running an MCP Server

# Start an MCP server
fort mcp-server --name "My MCP Server" --port 8888

# Create a custom sampling handler
fort mcp-handler my_handler --type sampling

# Create a tool handler
fort mcp-handler my_tool --type tool

# Create a resource handler
fort mcp-handler my_resource --type resource

Using MCP with Models

# Generate an MCP client for a model
fort model-mcp-client User --endpoint http://localhost:8888/mcp

# Generate an MCP server for a model
fort model-mcp-server User --port 8889

Sampling from LLMs

# Sample from LLM using MCP
result = await endpoint.sample_llm(
    prompt="What insights can you provide about this data?",
    system_prompt="You are a data analysis expert",
    max_tokens=1000
)

The client will display the sample request and handle user approval through the MCP protocol.

Accessing the UI

Once the servers are running:

• UI Server: http://localhost:9996
• API Server: http://localhost:9997
• API Documentation: http://localhost:9997/docs
• MCP Server: http://localhost:8888/mcp (when running)

Registry Integration

Endpoints are automatically registered as tools in the external registry at https://arthurcolle--registry.modal.run, making them available for agents to use.

License

MIT