Back to evaluations
Draft evaluation
Account Aggregator Credit Underwriting Multi-Agent System using AutoGen — evaluation
Evaluates multi-agent conversation history and final JSON loan decision payload.
Evaluation type
task based
Challenge
Account Aggregator Credit Underwriting Multi-Agent System using AutoGen
Difficulty
Intermediate
Rigor
Not declared
The author has not specified a rigor level.
Evaluation overview
How the linked challenge is judged: tasks, benchmarks, and criteria count.
Tasks
1
Benchmarks
0
Criteria
0
Task templates
Inputs and expected outputs.
Task 1
aa_credit_underwriting_eval
Validates financial input statement and returns structured credit recommendation.
Input format
JSON object with applicant_id, monthly_income, debt_obligations, and aa_bank_statement array.
Output format
JSON object with recommendation ('APPROVED', 'REJECTED', 'MANUAL_REVIEW'), max_eligible_loan, and compliance_score.