Green Hydrogen Feasibility Assessment Crew with CrewAI and Amazon Bedrock
Volatile renewable energy pricing and off-taker commitment shifts require rapid re-evaluation of green hydrogen project economics. Deploy a CrewAI multi-agent team running on Amazon Bedrock to evaluate electrolyzer capacity, solar tariffs, and off-take terms, outputting dynamic IRR models that accurately flag unviable project conditions.
What you are building
The core problem, expected build, and operating context for this challenge.
Build an autonomous multi-agent economic analysis team using CrewAI and Amazon Bedrock to evaluate green hydrogen megaproject feasibility.
How work is evaluated
Evaluates LCOH calculations and feasibility recommendations generated by the crew.
Challenge datasets
Public datasets, plus the private datasets you have access to.
How submissions are scored
These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.
correct_shelve_decision
Agent crew must recommend SHELVE when LCOH exceeds off-take cap
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
lcoh_calculation_accuracy
Accuracy of financial LCOH estimation relative to standard benchmark formula • target: 0.95 • range: 0-1
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
What you should walk away with
Define specialized CrewAI agents (Solar Analyst, Electrolyzer Specialist, Financial Modeler)
Configure Amazon Bedrock API provider using Anthropic Claude models
Implement sequential agent delegation task workflows for project IRR calculations
Identify tipping points where hydrogen production costs exceed market off-taker caps
Reference links and supporting material
Masdar Abu Dhabi green hydrogen project parameters, electrolyzer cost curves, and solar PPA pricing models.
Challenge setup
Evaluates LCOH calculations and feasibility recommendations generated by the crew.
Challenge input
JSON containing solar PPA tariff ($/MWh), electrolyzer cost ($/kW), and off-take cap ($/kg).
CrewAI
Required multi-agent role-playing framework
Amazon Bedrock
Required cloud model runtime environment
Evaluated output
JSON containing estimated LCOH ($/kg), project IRR %, and recommendation ('PROCEED' or 'SHELVE').
- Agent crew must recommend SHELVE when LCOH exceeds off-take cap
- Accuracy of financial LCOH estimation relative to standard benchmark formula • target: 0.95 • range: 0-1
- Benchmark: Renewable Energy Economic Analysis Benchmark
- Lcoh Calculation Accuracy target: 0.95
- 1 public reference case
- 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
- CrewAI · Required
- crewAI · Optional
- Amazon Bedrock · Optional
Evaluation contract
- correct_shelve_decision · Weight 1
- lcoh_calculation_accuracy · Weight 1
Metrics
- lcoh_calculation_accuracy · Target 0.95
Benchmarks
- Renewable Energy Economic Analysis Benchmark
- energy
Reference cases
- 1 public reference case
Workflow tasks
- Task 1
Input: JSON containing solar PPA tariff ($/MWh), electrolyzer cost ($/kW), and off-take cap ($/kg).
Output: JSON containing estimated LCOH ($/kg), project IRR %, and recommendation ('PROCEED' or 'SHELVE').
Execution harness: Python
Evaluation tools
- CrewAI · Agent Framework
- Amazon Bedrock · Inference Infrastructure
Recipe freeze marker
No recipe freeze time is recorded.