<<<<<<< HEAD

Supply Chain Disruption Response (OpenEnv)

A real-world-inspired OpenEnv environment for supply chain logistics and disruption response. The agent acts as a Lead Logistics Provider managing a supplier network, customer orders, and alternate sourcing choices under unpredictable market shocks.

Scenario

• The environment starts with a disruption: supplier downtime, port blockage, bankruptcy, or fuel shock.
• The agent must fulfill customer orders while balancing budget, delivery deadlines, alternate supplier capacity, and reputation.
• Observations include supplier network capabilities, pending order status, alternative sourcing options, budgets, and feedback signals.

Features

• OpenEnv-compliant SupplyChainEnv with reset(), step(), state()
• Strong type-safety via Pydantic models in models.py
• Rich observation space with supplier network, pending order status, and alternate sourcing details
• Dense reward shaped by delivery performance, budget discipline, negotiation, and reputation
• 7 action types for strategic disruption response decisions
• 3 grader tasks with easy → medium → hard difficulty progression
• FastAPI web service compatible with Hugging Face Spaces
• Docker validation script for quick integration testing

Action Space

• action_type: reroute_order, negotiate, split_order, notify_customer, expedite, cancel_order, query
• target_id: order or supplier identifier, e.g. O1 or Alt1
• value: numeric intensity for the action
• parameters: optional metadata such as supplier_id or supplier_ids

Observation Space

• supplier_network: available suppliers, capacity, cost, reliability, and lead time
• inventory_levels: current stock for Widget and Gadget
• pending_orders: customer orders with deadlines, penalty rates, assigned supplier, and delivery status
• disruption_event: active disruption type, severity, recovery timeline, and affected entity
• alternative_suppliers: backup source options for rerouting and split ordering
• budget_remaining: current cash available for sourcing, expedited delivery, and negotiation
• feedback: operational summary and customer communication status

Reward Structure

• value: dense trajectory reward
• on_time_deliveries: positive signal for deliveries meeting due dates
• budget_spent: cost pressure penalty
• carbon_penalty: environmental impact penalty
• pending_penalty: overdue shipment penalty
• offset_benefit: credit offset reward
• resilience_delta: reward for strengthening the supply chain
• reputation_delta: reward for improving customer reputation

Novel Features

• Supplier Network Resilience: manage alternate providers, negotiations, and split sourcing
• Customer Communication: notify customers proactively for late or at-risk orders
• Dynamic Disruption Types: support supplier outages, port blockages, bankruptcy, and fuel shocks
• Compound action space: route changes, negotiation, split sourcing, expedited delivery, cancellation, and intelligence queries
• Budget and penalty balancing: trade cost, delay, and reputational outcomes
• Docker validation script: docker-test.sh builds, runs, queries endpoints, and cleans up

Setup

git clone <this-repo> .
cd d:/supply_chain_openenv
python -m venv .venv
.venv/Scripts/activate
pip install -r requirements.txt

Run baseline

python baseline.py

Run as API

uvicorn main:app --host 0.0.0.0 --port 8000

Docker validation

chmod +x docker-test.sh
./docker-test.sh

OpenEnv metadata

• openenv.yaml contains environment metadata and action/observation schema

Task grader usage

Task summaries

• EasyTask: handle a single supplier outage by rerouting a disrupted order to meet its deadline while conserving budget.
• MediumTask: manage a port blockage with multiple orders, notify affected customers for late shipments, and preserve on-time service.
• HardTask: survive supplier bankruptcy or fuel shock through negotiation, split sourcing, and minimizing late penalties.

from tasks import EasyTask, MediumTask, HardTask
from engine import SupplyChainEnv
from models import Action

env = SupplyChainEnv(random_seed=42)
obs = env.reset()
action = Action(action_type='reroute_order', target_id='O1', value=1.0, parameters={'supplier_id': 'Alt1'})
obs, reward, done, info = env.step(action)

print('easy', EasyTask.evaluate(env, action))
print('medium', MediumTask.evaluate(env, action))
print('hard', HardTask.evaluate(env))

=======

SuppleChain

55e89601e7cad36de4f68a1cee21aee6e0e51849