Challenge

Graph-Based Identity Security Agent

This challenge focuses on building an intelligent agent system capable of analyzing complex user, system, and access relationships to identify security risks. You will construct a knowledge graph representing an enterprise's identity landscape, then leverage Gemini 3 for advanced reasoning over this graph. Your primary task is to develop a graph-based agent workflow using LangGraph to traverse and query this knowledge graph. The agent will integrate with LlamaIndex for Retrieval-Augmented Generation (RAG) to fetch relevant context from the graph, and utilize MCP-enabled tools to interact with simulated enterprise identity systems (e.g., Active Directory, HR databases). The agent should be able to perform extended thinking with adaptive reasoning budgets to detect anomalous access patterns, privilege escalations, and provide risk assessments.

Special Purpose AgentsHosted by Vera
Challenge brief

What you are building

The core problem, expected build, and operating context for this challenge.

This challenge focuses on building an intelligent agent system capable of analyzing complex user, system, and access relationships to identify security risks. You will construct a knowledge graph representing an enterprise's identity landscape, then leverage Gemini 3 for advanced reasoning over this graph. Your primary task is to develop a graph-based agent workflow using LangGraph to traverse and query this knowledge graph. The agent will integrate with LlamaIndex for Retrieval-Augmented Generation (RAG) to fetch relevant context from the graph, and utilize MCP-enabled tools to interact with simulated enterprise identity systems (e.g., Active Directory, HR databases). The agent should be able to perform extended thinking with adaptive reasoning budgets to detect anomalous access patterns, privilege escalations, and provide risk assessments.

Datasets

Shared data for this challenge

Review public datasets and any private uploads tied to your build.

Loading datasets...
Learning goals

What you should walk away with

  • Master LlamaIndex for building and querying a knowledge graph representing user, system, and access relationships within an enterprise.

  • Implement LangGraph to define stateful, cyclic agent workflows that navigate and analyze the identity knowledge graph.

  • Integrate Gemini 3 for sophisticated pattern recognition, contextual understanding, and risk assessment by providing it with retrieved graph data via RAG.

  • Design MCP server-side implementations for simulated identity management systems (e.g., user directories, access control lists) and create agent tools to interact with them.

  • Develop strategies for extended thinking within LangGraph, allowing the agent to dynamically allocate reasoning budget based on the complexity of the security query.

  • Build a RAG pipeline using LlamaIndex to retrieve relevant nodes and edges from the knowledge graph based on agent queries, feeding this context to Gemini 3.

How this agent runs

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 mitigatio...

Preview configuration

Challenge input

{'knowledge_graph_data': {'nodes': [{'id': 'user_alice', 'type': 'user', 'roles': ['admin']}, {'id': 'server_db', 'type': 'system', 'security_level...

Agent execution

The configured agent processes the input under the challenge policy.

Evaluated output

{'identified_risks': [{'risk_type': 'DormantAdmin', 'description': 'User Bob has admin privileges but no recent activity.', 'mitigation': 'Revoke p...

Checks for
  • At least 80% of actual hidden risks in the input graph are correctly identified.
Proof of success
  • MitigationQuality target: 85
Runtime evidence
  • Python execution harness
View technical recipe

Configured tools

No tool records are attached.

Evaluation contract

  • The evaluation module defines the checks.

Recipe state

This is a preview. The configuration can change before the evaluation recipe is locked.

Run this agent on your dataset and AI stack

Bring your dataset, model providers, and success criteria. We will scope the right managed run for your team.

Scope a managed run
Start from your terminal
$npx -y @versalist/cli start graph-based-identity-security-agent

[ok] Wrote CHALLENGE.md

[ok] Wrote .versalist.json

[ok] Wrote eval/examples.json

Requires VERSALIST_API_KEY. Works with any MCP-aware editor.

Docs
Manage API keys
Explore

Find another challenge

Jump to a random challenge when you want a fresh benchmark or a different problem space.

Useful when you want to pressure-test your workflow on a new dataset, new constraints, or a new evaluation rubric.

Frequently Asked Questions about Graph-Based Identity Security Agent