stateset-embedded

Local-first embedded commerce library for Python, powered by Rust.

Installation

pip install stateset-embedded

Supported environments: CPython 3.9-3.13. Prebuilt wheels ship for Linux x64/aarch64 (manylinux), macOS Intel/Apple Silicon, and Windows x64; everything else (including older glibc systems) builds from the sdist automatically wherever a Rust toolchain is available.

Optional framework adapters:

pip install "stateset-embedded[langchain]"
pip install "stateset-embedded[crewai]"
pip install "stateset-embedded[autogen]"
# or install all Python framework helpers together
pip install "stateset-embedded[agents]"

Or build from source:

cd bindings/python
pip install maturin
maturin develop

Quick Start

from stateset_embedded import Commerce, CreateOrderItemInput

# Initialize with SQLite database
commerce = Commerce("./store.db")

# Or use in-memory database for testing
commerce = Commerce(":memory:")

# Create a customer
customer = commerce.customers.create(
    email="alice@example.com",
    first_name="Alice",
    last_name="Smith"
)
print(f"Created customer: {customer.id}")

# Create a product with variant
product = commerce.products.create(
    name="Premium Widget",
    description="A high-quality widget"
)

# Create inventory
item = commerce.inventory.create_item(
    sku="WIDGET-001",
    name="Premium Widget",
    initial_quantity=100
)

# Check stock
stock = commerce.inventory.get_stock("WIDGET-001")
print(f"Available: {stock.total_available}")

# Create an order
order = commerce.orders.create(
    customer_id=customer.id,
    items=[
        CreateOrderItemInput(
            sku="WIDGET-001",
            name="Premium Widget",
            quantity=2,
            unit_price=29.99
        )
    ]
)
print(f"Order {order.order_number}: ${order.total_amount}")

# Ship the order
commerce.orders.ship(order.id, tracking_number="1Z999AA10123456784")

# Analytics
summary = commerce.analytics.sales_summary(period="last30days")
print(f"Revenue: ${summary.total_revenue}")

# Currency conversion (set a rate, then convert)
commerce.currency.set_rate("USD", "EUR", 0.92, source="manual")
conversion = commerce.currency.convert("USD", "EUR", 100.0)
print(f"$100 USD = €{conversion.converted_amount} EUR")

Agent Toolkit

The Python package also ships a native agent toolkit for core embedded commerce operations:

from stateset_embedded import Commerce, create_embedded_agent_toolkit

commerce = Commerce(":memory:")
toolkit = create_embedded_agent_toolkit(commerce, allow_apply=False)

openai_tools = toolkit.get_tools(format="openai")
descriptors = toolkit.create_tool_descriptors(
    filter=["list_customers", "list_orders", "get_sales_summary"]
)
callable_registry = toolkit.create_callable_registry(filter=["list_customers"])
langchain_tools = toolkit.create_langchain_tools(filter=["list_customers"])

execution = toolkit.execute_openai_tool_call(
    {
        "call_id": "py_demo_1",
        "function": {
            "name": "list_customers",
            "arguments": "{\"limit\": 5}",
        },
    }
)
print(execution["output_message"])

This toolkit is aimed at Python agent runtimes such as CrewAI- or AutoGen-style hosts that need core commerce operations in-process. Use the JS toolkit or MCP server when you need the full registry-generated tool surface and policy runtime.

OpenAI-compatible helper methods are available out of the box, and the framework-specific helpers are available when the corresponding framework packages are installed:

• create_tool_descriptors()
• create_callable_registry()
• execute_tool() and execute_tool_calls()
• create_openai_tools()
• create_langchain_tools()
• create_crewai_tools()
• create_autogen_tools()

Each helper also accepts a tool_factory callback so you can generate your own framework objects from the native descriptors.

The package also exposes helper modules for direct imports:

from stateset_embedded.generic import create_tool_descriptors, create_callable_registry
from stateset_embedded.openai import create_openai_tools, execute_openai_tool_call
from stateset_embedded.langchain import create_langchain_tools
from stateset_embedded.crewai import create_crewai_tools
from stateset_embedded.autogen import create_autogen_tools

descriptors = create_tool_descriptors(commerce, filter=["list_customers"])
registry = create_callable_registry(commerce, filter=["list_customers"])
openai_tools = create_openai_tools(commerce, filter=["list_customers"])
langchain_tools = create_langchain_tools(commerce, filter=["list_customers"])
crewai_tools = create_crewai_tools(commerce, filter=["count_customers"])
autogen_tools = create_autogen_tools(commerce, filter=["get_sales_summary"])

Runnable repo examples for those module imports live in:

• examples/python/openai_tools.py
• examples/python/generic_tools.py
• examples/python/langchain_tools.py
• examples/python/crewai_tools.py
• examples/python/autogen_tools.py
• examples/python/framework_adapters.py

Sequencer Sync

from stateset_embedded import SyncRuntime
import json

runtime = SyncRuntime(json.dumps({
    "sequencer_base_url": "http://127.0.0.1:4000",
    "engine": {
        "agent_id": "agent-1",
        "tenant_id": "tenant-1",
        "store_id": "store-1",
        "outbox_path": "/tmp/stateset-sync-outbox.json",
        "state_path": "/tmp/stateset-sync-state.json"
    },
    "agent_key_id": 7
}))

runtime.record(
    "order.created",
    "order",
    "ORD-1001",
    json.dumps({"total": 42.50}),
    command_id="cmd-1001",
)

push = runtime.push()
status = runtime.status()
print(push.remote_head)
print(status.caught_up)

SyncRuntime mirrors the Rust SDK sync surface for local event recording, sequencer health checks, remote-head refresh, push/pull/full-sync operations, and inspection of confirmations, dead letters, and buffered pulled events. Typed Python classes are available for the main sync surfaces, and the JSON helpers remain available when a serialized view is more convenient.

Features

• Local-First: All data stored in SQLite, works offline
• Zero Dependencies: Single native extension, no external services
• Type Safe: Full type hints and IDE support
• Fast: Native Rust performance
• Analytics + Currency: Built-in reporting/forecasting and multi-currency operations
• Vector Search: Semantic search for products/customers (opt-in, OpenAI embeddings)

API Reference

Commerce

Main entry point for all operations.

commerce = Commerce("./store.db")  # SQLite file
commerce = Commerce(":memory:")     # In-memory database

Customers

# Create
customer = commerce.customers.create(
    email="alice@example.com",
    first_name="Alice",
    last_name="Smith",
    phone="+1234567890",
    accepts_marketing=True
)

# Get by ID or email
customer = commerce.customers.get(customer_id)
customer = commerce.customers.get_by_email("alice@example.com")

# List all
customers = commerce.customers.list()

# Count
count = commerce.customers.count()

Orders

# Create
order = commerce.orders.create(
    customer_id=customer.id,
    items=[
        CreateOrderItemInput(
            sku="SKU-001",
            name="Product Name",
            quantity=2,
            unit_price=29.99
        )
    ],
    currency="USD",
    notes="Gift wrap please"
)

# Get
order = commerce.orders.get(order_id)

# List all
orders = commerce.orders.list()

# Update status
order = commerce.orders.update_status(order_id, "processing")

# Ship with tracking
order = commerce.orders.ship(order_id, tracking_number="1Z123...")

# Cancel
order = commerce.orders.cancel(order_id)

Products

# Create with variants
from stateset_embedded import CreateProductVariantInput

product = commerce.products.create(
    name="Premium Widget",
    description="High-quality widget",
    variants=[
        CreateProductVariantInput(
            sku="WIDGET-SM",
            price=19.99,
            name="Small"
        ),
        CreateProductVariantInput(
            sku="WIDGET-LG",
            price=29.99,
            name="Large"
        )
    ]
)

# Get by ID
product = commerce.products.get(product_id)

# Get variant by SKU
variant = commerce.products.get_variant_by_sku("WIDGET-SM")

# List all
products = commerce.products.list()

Inventory

# Create inventory item
item = commerce.inventory.create_item(
    sku="WIDGET-001",
    name="Premium Widget",
    description="High-quality widget",
    initial_quantity=100,
    reorder_point=10
)

# Check stock levels
stock = commerce.inventory.get_stock("WIDGET-001")
print(f"On hand: {stock.total_on_hand}")
print(f"Allocated: {stock.total_allocated}")
print(f"Available: {stock.total_available}")

# Adjust stock
commerce.inventory.adjust("WIDGET-001", -5, "Sold 5 units")
commerce.inventory.adjust("WIDGET-001", 50, "Received shipment")

# Reserve for order
reservation = commerce.inventory.reserve(
    sku="WIDGET-001",
    quantity=2,
    reference_type="order",
    reference_id=order_id,
    expires_in_seconds=3600  # 1 hour
)

# Confirm reservation (deducts from on-hand)
commerce.inventory.confirm_reservation(reservation.id)

# Or release reservation (returns to available)
commerce.inventory.release_reservation(reservation.id)

Vector Search

# Initialize vector search (requires OpenAI API key)
vector = commerce.vector("sk-...")

# Index a product or customer
vector.index_product(product.id)

# Semantic search
results = vector.search_products("wireless bluetooth headphones", limit=10)
for r in results:
    print(r.id, r.score, r.name)

# Stats/maintenance
stats = vector.stats()
vector.clear("products")

Returns

from stateset_embedded import CreateReturnItemInput

# Create return request
ret = commerce.returns.create(
    order_id=order.id,
    reason="defective",
    items=[
        CreateReturnItemInput(
            order_item_id=order.items[0].id,
            quantity=1
        )
    ],
    reason_details="Product arrived damaged"
)

# Get return
ret = commerce.returns.get(return_id)

# Approve return
ret = commerce.returns.approve(return_id)

# Reject return
ret = commerce.returns.reject(return_id, "Item was used")

# List all returns
returns = commerce.returns.list()

Carts / Checkout

from stateset_embedded import CartAddress, AddCartItemInput

# Create a cart (guest checkout)
cart = commerce.carts.create(customer_email="alice@example.com", currency="USD")

# Add items
commerce.carts.add_item(
    cart_id=cart.id,
    item=AddCartItemInput(
        sku="SKU-001",
        name="Widget",
        quantity=2,
        unit_price=29.99,
    ),
)

# Set shipping (address + selection)
address = CartAddress(
    first_name="Alice",
    last_name="Smith",
    line1="123 Main St",
    city="San Francisco",
    postal_code="94105",
    country="US",
)
cart = commerce.carts.set_shipping(
    cart.id,
    address,
    shipping_method="standard",
    shipping_carrier="ups",
    shipping_amount=9.99,
)

# Reserve inventory for cart items (optional)
cart = commerce.carts.reserve_inventory(cart.id)

# Complete checkout (creates an order)
result = commerce.carts.complete(cart.id)
print(result.order_number)

Analytics

# Sales summary
summary = commerce.analytics.sales_summary(period="last30days")
print(summary.total_revenue, summary.order_count)

# Top products / customers
top_products = commerce.analytics.top_products(period="this_month", limit=10)
top_customers = commerce.analytics.top_customers(period="all_time", limit=10)

# Forecasting
forecasts = commerce.analytics.demand_forecast(days_ahead=30)
revenue = commerce.analytics.revenue_forecast(periods_ahead=3, granularity="month")

Currency

# Set an exchange rate
commerce.currency.set_rate("USD", "EUR", 0.92, source="manual")

# Convert currency
conversion = commerce.currency.convert("USD", "EUR", 100.0)
print(conversion.converted_amount)

# Store settings
settings = commerce.currency.get_settings()
settings = commerce.currency.enable_currencies(["USD", "EUR", "GBP"])

Order Statuses

• pending - Order created, awaiting confirmation
• confirmed - Order confirmed
• processing - Order being processed
• shipped - Order shipped
• delivered - Order delivered
• cancelled - Order cancelled
• refunded - Order refunded

Return Reasons

• defective - Product is defective
• not_as_described - Product not as described
• wrong_item - Wrong item received
• no_longer_needed - No longer needed
• changed_mind - Changed mind
• better_price_found - Found better price elsewhere
• damaged - Product arrived damaged
• other - Other reason

Development

# Install dev dependencies
pip install maturin pytest

# Build in development mode
maturin develop

# Run tests
pytest tests/

# Build release wheel
maturin build --release

License

MIT OR Apache-2.0