Skip to main content
Loading navigation
Back to challenges

Industrial Safety Network Switch Failure Root-Cause Analysis Agent with Claude Agents SDK

Unannounced industrial network switch failures can disrupt safety-critical interlocks and blind plant operators. Build a forensic diagnostic agent using Claude Agents SDK to correlate packet loss timelines with PLC safety trip logs, pinpointing root-cause network faults with zero false positives.

EngineeringHosted by Vera
Sign in free to start

What you are building

Develop an autonomous forensic investigation agent using Claude Agents SDK to analyze safety-critical network switch failures and control logic degradation.

How work is evaluated

Evaluation

Evaluates correctness of network switch failure root cause pinpointing and incident safety analysis.

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

Fault Isolation Test

Confirms switch port physical link failure is isolated.

Binary check

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

Dimension 2

false_positive_rate

Rate of misidentified network anomalies. • target: 0 • 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

  • Implement Claude Agents SDK workflows to perform timeline correlation across network and safety PLC events

  • Analyze network switch PCAP metadata alongside SIL-3 rated PLC interlock event buffers

  • Synthesize formal safety incident root-cause reports compliant with IEC 61508 / IEC 61511 standards

Resources

Dataset notes

Correlated syslog datasets from managed Ethernet switches and SIL-3 Safety Instrumented System (SIS) controllers during simulated link failure events.

Challenge setup

Evaluates correctness of network switch failure root cause pinpointing and incident safety analysis.

Challenge input

JSON with switch_event_log, sis_trip_log, and ptp_clock_skew_ns

Claude Agents SDK

Supports extended thinking and tool selection for complex forensic reasoning.

Evaluated output

JSON with primary_fault_source, triggered_sis_interlock, and false_positive_flag

Evaluation checks
  • Confirms switch port physical link failure is isolated.
  • Rate of misidentified network anomalies. • target: 0 • range: 0-1
Configured evaluation evidence
  • Benchmark: Industrial Cyber-Safety Forensics Benchmark
  • False Positive Rate target: 0%
  • 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
  • Claude Code · Required
Observation
  • Claude Code · Required
Policy Serving
  • Claude 4.1 Opus · Optional
  • Claude 4 Sonnet · Optional

Evaluation contract

  • Fault Isolation Test · Weight 1
  • false_positive_rate · Weight 1

Metrics

  • false_positive_rate · Target 0

Benchmarks

  • Industrial Cyber-Safety Forensics Benchmark
  • security

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON with switch_event_log, sis_trip_log, and ptp_clock_skew_ns

    Output: JSON with primary_fault_source, triggered_sis_interlock, and false_positive_flag

    Execution harness: Python

Evaluation tools

  • Claude Agents SDK · Agent Core Framework

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Industrial Safety Network Switch Failure Root-Cause Analysis Agent with Claude Agents SDK