Skip to main content
Loading navigation
Back to challenges

Google ADK ISA/IEC 62443 Security Compliance Agent with Patronus AI

Adopting ISA/IEC 62443 cybersecurity standards requires continuous auditing of OT network configurations and safety control logic. Construct a multi-modal security auditing agent using Google ADK and Patronus AI guardrails to evaluate industrial switch configs and network traffic logs. Success is measured by detecting 100% of policy violations while maintaining zero false positives on compliant configurations.

EngineeringHosted by Vera
Sign in free to start

What you are building

Build a Google ADK agent with Gemini models and Patronus AI evaluation frameworks to continuously audit OT network configs against ISA/IEC 62443 benchmarks.

How work is evaluated

Evaluation

Evaluates detection of cybersecurity policy violations in network switch and PLC firewall configurations.

Datasets

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

How submissions are scored

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

detect_telnet_usage

Flags telnet/unencrypted plaintext protocols as non-compliant.

Binary check

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

Dimension 2

audit_recall

Recall metric for detecting known non-compliant configurations. • 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

  • Initialize Google Agent Development Kit (ADK) agent pipeline targeting Gemini models.

  • Implement Patronus AI evaluators for ISA/IEC 62443-4-2 cybersecurity requirements.

  • Analyze network switch configuration files and identify unencrypted industrial protocols.

  • Generate actionable remediation reports for OT system administrators.

Resources

Dataset notes

Network switch configuration files and firewall rulesets collected from synthetic substation and industrial plant environments.

Challenge setup

Evaluates detection of cybersecurity policy violations in network switch and PLC firewall configurations.

Challenge input

JSON object with device_id and config_text

Google ADK

Required framework for building Gemini-powered agents.

Patronus AI

Provides policy-level evaluation and guardrail checks.

Evaluated output

JSON object with compliant_flag, violations, and patronus_eval_score

Evaluation checks
  • Flags telnet/unencrypted plaintext protocols as non-compliant.
  • Recall metric for detecting known non-compliant configurations. • target: 1 • range: 0-1
Configured evaluation evidence
  • Benchmark: OTCyberBench-62443
  • Audit Recall target: 1
  • 1 public reference case
Configured runtime
  • Python execution harness
  • Python sandbox (unavailable on Versalist)

Shows the first task, up to three tools, and up to three evidence items.

View technical recipe

Configured tools

Action Space
  • Google · Optional
  • Google ADK · Optional
Reward / Eval
  • Patronus AI · Required
Safety / Guardrails
  • Patronus AI · Required

Evaluation contract

  • detect_telnet_usage · Weight 1
  • audit_recall · Weight 1

Metrics

  • audit_recall · Target 1

Benchmarks

  • OTCyberBench-62443
  • security

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON object with device_id and config_text

    Output: JSON object with compliant_flag, violations, and patronus_eval_score

    Execution harness: Python

Evaluation tools

  • Google ADK · Agent Framework
  • Patronus AI · AI Workflow Automation

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Google ADK ISA/IEC 62443 Security Compliance Agent with Patronus AI