Generate an RL-optimized drone deployment and evacuation strategy.

Uses reinforcement learning models trained on thousands of simulated disasters to recommend optimal resource allocation, routing, and risk parameters for a specific incident or pre-alert.

Args: incident_id: Incident identifier (INC-XXX) or pre-alert ID (PRE-XXX) to generate strategy for.

Returns: OptimizationStrategy: Complete strategy recommendation with: - strategy_id: Unique identifier - related_alert_id: Original incident/alert ID - recommended_deployment: Drone type counts - evacuation_routes: Prioritized route list - estimated_success_rate: Predicted success (0.0-1.0) - simulation_proof_url: NeoFS evidence link

Example: >>> strategy = await get_deployment_strategy("PRE-ABC123") >>> print(strategy.strategy_id) >>> print(strategy.recommended_deployment) # {"surveillance": 2, ...} >>> print(f"Success rate: {strategy.estimated_success_rate:.0%}")

Use Cases: - Pre-positioning drones before predicted disasters (PDIE alerts) - Active response optimization for confirmed incidents - Multi-objective optimization (speed, safety, resource efficiency) - Scenario comparison and sensitivity analysis

Integration: Strategy linked to blockchain for immutable audit trail. After approval, use update_mission_params to push to drones.

Trigger a Digital Twin physics simulation for disaster scenario modeling.

Queues a high-fidelity simulation job and returns immediately with a job ID. Clients should subscribe to the simulation resource URI for real-time progress updates and result notification.

Workflow: 1. Validate simulation request parameters 2. Generate unique simulation ID 3. Queue job to DTSOP backend (Unity/Unreal Engine) 4. Store job metadata in simulation registry 5. Return simulation ID and subscription URI 6. Background processor updates status → processing → completed 7. Client fetches results from NeoFS when completed

Args: request: SimulationRequest with: - scenario_id: Unique scenario identifier - sector_id: Geographic sector to simulate - disaster_type: Physics model (flood/wildfire/earthquake) - parameters: Scenario params (wind_speed, water_level, etc.) - priority: "standard" or "urgent" ctx: Optional FastMCP context for logging.

Returns: str: Message with simulation ID and subscription instructions: "Simulation queued with ID: SIM-XXXXXXXX. Subscribe to resq://simulations/SIM-XXXXXXXX for updates."

Example: >>> from resq_mcp.models import SimulationRequest >>> request = SimulationRequest( ... scenario_id="flood-001", ... sector_id="Sector-1", ... disaster_type="flood", ... parameters={"water_level": 2.5}, ... priority="urgent" ... ) >>> result = await run_simulation(request) >>> print(result) # "Simulation queued with ID: SIM-ABCD1234..."

Integration: Production would: - Validate request against simulation templates - Check cluster capacity and queue position - Store job in Redis with priority - Submit to Unity/Unreal Engine processing cluster - Return estimated completion time

Submit validation result for an incident report.

Used by human operators or automated validation systems (HCE) to confirm or reject incident reports before triggering full response.

Args: val: IncidentValidation with: - incident_id: ID of incident being validated - is_confirmed: True=confirmed, False=rejected/false positive - validation_source: Who/what validated (e.g., "Human-Operator") - correlated_pre_alert_id: Optional linked PDIE alert - notes: Validation reasoning and evidence

Returns: str: Confirmation message indicating action taken: "Incident {id} successfully CONFIRMED." or "Incident {id} successfully REJECTED."

Example: >>> from resq_mcp.models import IncidentValidation >>> validation = IncidentValidation( ... incident_id="INC-123", ... is_confirmed=True, ... validation_source="Human-Operator-Alice", ... notes="Confirmed via video evidence and ground reports" ... ) >>> result = await validate_incident(validation) >>> print(result) # "Incident INC-123 successfully CONFIRMED."

Workflow: 1. Edge AI detects incident (low confidence) 2. HCE cross-references with PDIE/sensors 3. If ambiguous → human review required 4. Operator submits validation via this tool 5. If confirmed → trigger response strategy 6. If rejected → log as false positive, update ML model

Audit Trail: All validations logged with timestamp, source, and reasoning for post-incident analysis and ML model refinement.