Skip to main content
Back to challenges
Challenge

Water Utility Physical Cyber Attack Monitoring with Pydantic AI

Cyber-physical attacks on water treatment plants often cause physical operational drift without triggering network-level compromises. Build a type-safe process physics monitoring agent using Pydantic AI to detect unauthorized flow rate manipulations and achieve 100% type-safe validation.

EngineeringHosted by Vera
Sign in free to start
Challenge brief

What you are building

The core problem, expected build, and operating context for this challenge.

Develop a type-safe OT process validation agent using Pydantic AI to flag anomalous flow physics without explicit network breaches.

Delivery guide

How work is evaluated

Evaluation

Evaluates Pydantic AI model's ability to catch physical law violations in water pipeline telemetry.

Datasets

Challenge datasets

Public datasets, plus the private datasets you have access to.

Evaluation rubric

How submissions are scored

These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.

Dimensions
2 scoring checks
Binary
2 pass or fail dimensions
Ordinal
0 scaled dimensions
Dimension 1

pydantic_schema_valid

Verifies return payload strictly adheres to Pydantic output model.

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Dimension 2

violation_detection_accuracy

Accuracy in detecting physical conservation law anomalies • target: 1 • range: 0-1

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Learning goals

What you should walk away with

  • Define Pydantic models for physical fluid dynamic states (flow rate, valve position, pressure)

  • Build a Pydantic AI agent with structured dependencies and input validation

  • Identify non-physical telemetry states indicative of direct physical tampering

  • Emit typed alert models for plant SCADA operator intervention

Resources and assets

Reference links and supporting material

Dataset notes

Water network telemetry logs reflecting hydraulic flow measurements and physical tampering scenarios.

Challenge setup

Evaluates Pydantic AI model's ability to catch physical law violations in water pipeline telemetry.

Challenge input

JSON object with flow_in_gpm, flow_out_gpm, valve_open_pct, pressure_psi

Pydantic AI

Type-safe Python framework for structured AI agent outputs

Evaluated output

JSON object with physical_violation (boolean), anomaly_type (string), certainty (float)

Evaluation checks
  • Verifies return payload strictly adheres to Pydantic output model.
  • Accuracy in detecting physical conservation law anomalies • target: 1 • range: 0-1
Configured evaluation evidence
  • Benchmark: Water-CyberPhysical-2025
  • Violation Detection Accuracy target: 1
  • 1 public reference case
Configured runtime
  • Python execution harness
  • Python sandbox (unavailable on Versalist)

Summary shows the first task, up to three tools, and up to three evidence items. Open the technical recipe to view all configured items.

View technical recipe

Configured tools

Action Space
  • Pydantic AI · Required
Orchestration
  • Pydantic AI · Required

Evaluation contract

  • pydantic_schema_valid · Weight 1
  • violation_detection_accuracy · Weight 1

Metrics

  • violation_detection_accuracy · Target 1

Benchmarks

  • Water-CyberPhysical-2025
  • security

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON object with flow_in_gpm, flow_out_gpm, valve_open_pct, pressure_psi

    Output: JSON object with physical_violation (boolean), anomaly_type (string), certainty (float)

    Execution harness: Python

Evaluation tools

  • Pydantic AI · agent_framework

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Water Utility Physical Cyber Attack Monitoring with Pydantic AI