Shift Handover Summary Agent with CrewAI
When control engineering and maintenance teams change shifts, critical ongoing fault investigations and machine status updates can be lost. In this challenge, developers will deploy a two-agent CrewAI system (Shift Summarizer Agent & Safety Inspector Agent) that reads continuous SCADA event logs and generates concise, safety-audited shift handover reports.
What you are building
Deploy a CrewAI shift handover agent team to summarize SCADA downtime events and safety alerts for incoming plant crews.
How work is evaluated
Evaluates accuracy and safety highlight inclusion in generated handover summaries.
Datasets
Public datasets, plus the private datasets you have access to.
How submissions are scored
Alarm Highlight Check
High temperature alert must be explicitly listed in critical alerts array
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
Log Summary Completeness
Percentage of downtime events captured accurately • target: 1 • range: 0-1
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
Learning goals
Define CrewAI agents with clear shift handover reporting responsibilities
Process raw SCADA downtime logs and produce chronologically ordered summaries
Highlight unresolved safety interlocks requiring immediate attention
Format output reports in standard industrial shift log structures
Resources
Synthetic 8-hour shift event logs containing machine runtime and fault events.
Challenge setup
Evaluates accuracy and safety highlight inclusion in generated handover summaries.
Challenge input
JSON list of SCADA events
CrewAI
Assigned framework
Evaluated output
JSON object with total_downtime_min, critical_alerts_highlighted (array)
- High temperature alert must be explicitly listed in critical alerts array
- Percentage of downtime events captured accurately • target: 1 • range: 0-1
- Benchmark: Industrial Text Summarization Benchmark
- Log Summary Completeness target: 1
- 1 public reference case
- 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
- CrewAI · Required
- crewAI · Optional
Evaluation contract
- Alarm Highlight Check · Weight 1
- Log Summary Completeness · Weight 1
Metrics
- Log Summary Completeness · Target 1
Benchmarks
- Industrial Text Summarization Benchmark
- general
Reference cases
- 1 public reference case
Workflow tasks
- Task 1
Input: JSON list of SCADA events
Output: JSON object with total_downtime_min, critical_alerts_highlighted (array)
Execution harness: Python
Evaluation tools
- CrewAI · Multi-agent framework
Recipe freeze marker
No recipe freeze time is recorded.