MCP AI Infrastructure
ReUp includes a sophisticated AI/MCP (Model Context Protocol) infrastructure for autonomous operations.
Overview
The MCP server exposes 28 tool modules that enable AI agents to interact with the platform:
┌─────────────────────────────────────────────────────────────┐
│ MCP Server Layer │
├─────────────────────────────────────────────────────────────┤
│ ┌─────────────┐ ┌─────────────┐ ┌─────────────┐ │
│ │ FastMCP │ │ Auth/RBAC │ │ Circuit │ │
│ │ Transport │ │ Middleware │ │ Breaker │ │
│ └─────────────┘ └─────────────┘ └─────────────┘ │
├─────────────────────────────────────────────────────────────┤
│ 28 Tool Modules │
├─────────────────────────────────────────────────────────────┤
│ Analytics │ Approvals │ Commerce │ Cultivation │ Dashboard │
│ Delivery │ Dutchie │ Imports │ Inventory │ Invoices │
│ LabTest │ METRC* │ ML │ Notify │ Onboard │
│ Orders │ Products │ Receive │ Relations │ Sync │
│ Workflows │ ToolSearch│ │
└─────────────────────────────────────────────────────────────┘
Tool Categories
Inventory & Products
| Tool | Purpose |
|---|---|
tools_inventory | Query inventory levels, availability |
tools_products | Product catalog operations |
tools_imports | Multi-source product import |
Orders & Commerce
| Tool | Purpose |
|---|---|
tools_orders | Order management operations |
tools_commerce | B2B marketplace operations |
tools_invoices | Invoice lifecycle management |
Compliance (METRC)
| Tool | Purpose |
|---|---|
tools_metrc | Core METRC operations |
tools_metrc_read | Read-only METRC data |
tools_metrc_actions | Queued write operations |
tools_metrc_executors | Action execution |
tools_metrc_locations | Facility/location data |
tools_metrc_write_extended | Extended write ops |
Analytics & Intelligence
| Tool | Purpose |
|---|---|
tools_analytics | Analytics queries, forecasts |
tools_dashboard | Dashboard metrics, KPIs |
tools_ml_infrastructure | ML model inference |
Operations
| Tool | Purpose |
|---|---|
tools_delivery | Delivery operations |
tools_cultivation | Cultivation data |
tools_receiving | Receiving workflows |
tools_workflows | Workflow orchestration |
tools_approvals | AI approval workflows |
tools_lab_testing | Lab testing data |
Integration & Sync
| Tool | Purpose |
|---|---|
tools_dutchie | Dutchie POS integration |
tools_sync | Data synchronization |
tools_onboarding | Onboarding workflows |
tools_relationships | Partnership management |
tools_notifications | Notification dispatch |
Discovery
| Tool | Purpose |
|---|---|
tools_tool_search | Find relevant tools for tasks |
Authentication & Access Control
Persona-Based Access
The MCP server uses persona-based authentication:
- Entity-scoped: Tools operate within entity boundaries
- Role-filtered: Tools filtered by user role
- Delegation: AI can act on behalf of users
Security Layers
- JWT authentication
- Entity context validation
- Role-based tool filtering
- Audit logging
ML Infrastructure (2026 Agentic Commerce)
Advanced ML capabilities:
| Service | Purpose |
|---|---|
autonomous_execution_service | Autonomous agent execution |
ml_orchestrator | ML pipeline management |
ensemble_forecasting | Multi-model forecasting |
calibration_service | Model calibration |
explainability_service | Model interpretability |
feature_store | Feature engineering |
outcome_tracker | Prediction tracking |
ab_testing_service | A/B testing framework |
Usage Examples
Tool Discovery
# Find tools for inventory management
result = await tools_tool_search.search("inventory levels low stock")
Inventory Query
# Get low stock items
result = await tools_inventory.get_low_stock(
entity_id="entity-uuid",
threshold=10
)
METRC Operations
# Read packages from METRC
packages = await tools_metrc_read.get_packages(
facility_license="C10-0000001",
active_only=True
)