Back to evaluations
Public evaluation

IdentityRiskAssessment

The agent will be evaluated on its ability to accurately identify pre-defined identity-related security risks within a given knowledge graph, provide clear explanations for the risks, and suggest appropriate mitigation strategies. Evaluation will also cover the efficiency of its graph traversal and reasoning process.

Evaluation type
task based
Challenge
Graph-Based Identity Security 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

IdentityRiskAssessment

Given a knowledge graph representing enterprise identities and access, identify specific security vulnerabilities (e.g., dormant accounts with high privileges, unauthorized access paths).

Input format

{'knowledge_graph_data': {'nodes': [{'id': 'user_alice', 'type': 'user', 'roles': ['admin']}, {'id': 'server_db', 'type': 'system', 'security_level': 'confidential'}], 'edges': [{'source': 'user_alice', 'target': 'server_db', 'relation': 'accesses'}], 'query': 'Identify dormant high-privilege accounts.'}}

Output format

{'identified_risks': [{'risk_type': 'DormantAdmin', 'description': 'User Bob has admin privileges but no recent activity.', 'mitigation': 'Revoke privileges or disable account'}], 'reasoning_path': ['...']}