Back to evaluations
Public evaluation
COSMIC AML Alert Analysis Task
Evaluates Pydantic AI models on structured validation accuracy, privacy leakage checks, and risk calculation accuracy.
Evaluation type
task based
Challenge
Pydantic AI & Dify Anti-Money Laundering Sharing Agent
Difficulty
Advanced
Rigor
Unspecified
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
COSMIC AML Alert Analysis Task
Parses transaction log and produces structured ML/TF risk assessment
Input format
JSON containing obfuscated transaction logs and counterparty metadata
Output format
JSON matching Pydantic output model with risk_level, confidence, and flags