Skip to main content
Back to challenges
Challenge

Agents for Dynamic Spatial Strategy Analysis

This challenge focuses on building a sophisticated multi-agent system capable of analyzing game clips to understand spatial reasoning and tactical decision-making. You will design and implement a graph-based workflow using LangGraph, where specialized agents communicate via an A2A protocol. These agents, powered by GPT-5 (or latest available OpenAI model) and Claude Opus 4.1 (or latest available Anthropic model), will collaborate to process video data, extract spatial information, identify key events, and synthesize strategic insights. The core of the system will involve robust tool integration, possibly using an MCP-like approach, to interact with hypothetical video processing APIs and external knowledge bases. The system will demonstrate extended thinking capabilities, allowing agents to iteratively refine their understanding of complex in-game scenarios. This involves dynamically adjusting reasoning budgets based on the complexity of the observed events and the uncertainty in the analysis. The goal is to produce actionable strategic feedback or predictive insights based on the observed spatial patterns and agent interactions within the game clips, pushing beyond simple event detection to true strategic understanding.

Special Purpose AgentsHosted by Vera
Sign in free to start
Challenge brief

What you are building

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

Build a graph-based multi-agent system using LangGraph with GPT-5 and Claude Opus 4.1 for autonomous spatial reasoning and strategy analysis from dynamic game clips.

Delivery guide

How work is evaluated

Evaluation

The evaluation will assess the system's ability to accurately analyze spatial reasoning in simulated game data, its adherence to LangGraph's graph-based workflow, effective A2A communication, and the utility of generated strategic insights. Performance will be judged on the quality of agent collaboration and the depth of analysis.

Datasets

Challenge datasets

Public datasets, plus the private datasets you have access to.

Learning goals

What you should walk away with

  • Design and implement stateful, graph-based agent workflows using LangGraph for complex, iterative analysis.

  • Synthesize strategic insights and generate actionable feedback based on multi-agent analysis of spatial reasoning.

  • Integrate external tools (e.g., hypothetical video analysis APIs) into the agentic workflow using an MCP-inspired pattern.

  • Develop extended thinking models with adaptive reasoning budgets to handle varying complexity in dynamic spatial data.

  • Establish robust A2A (Agent-to-Agent) communication protocols for seamless collaboration between specialized LLM agents.

Resources and assets

Reference links and supporting material

Dataset notes

Sample data for 2 tasks

Challenge setup

The evaluation will assess the system's ability to accurately analyze spatial reasoning in simulated game data, its adherence to LangGraph's graph-based workflow, effective A2A communication, and the utility of genera...

Challenge input

{'clip_id': 'string', 'frame_data': [{'frame_num': int, 'objects': [{'id': 'string', 'type': 'string', 'x': int, 'y': int}]}]}

Agent execution

No agent tool is listed for this challenge.

Evaluated output

{'clip_id': 'string', 'extracted_events': [{'type': 'string', 'timestamp': int, 'details': {}}]}

Evaluation checks
  • Ensure the solution uses a defined LangGraph DAG with distinct agent nodes and clear state transitions.
  • Verify that agents communicate explicitly via a defined A2A protocol (e.g., custom message formats, shared...
  • Confirm implementation of adaptive reasoning budgets or iterative refinement loops for complex scenarios.
Configured evaluation evidence
  • SpatialEventAccuracy target: 85
Configured runtime
  • Python execution harness

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

No tool records are attached.

Evaluation contract

  • Ensure the solution uses a defined LangGraph DAG with distinct agent nodes and clear state transitions.
  • Verify that agents communicate explicitly via a defined A2A protocol (e.g., custom message formats, shared state updates).
  • Confirm implementation of adaptive reasoning budgets or iterative refinement loops for complex scenarios.

Metrics

  • SpatialEventAccuracy · Target 85
  • StrategicInsightRelevance · Target 4
  • ComputationalEfficiency · Target 50000

Workflow tasks

  1. Task 1

    Input: {'clip_id': 'string', 'frame_data': [{'frame_num': int, 'objects': [{'id': 'string', 'type': 'string', 'x': int, 'y': int}]}]}

    Output: {'clip_id': 'string', 'extracted_events': [{'type': 'string', 'timestamp': int, 'details': {}}]}

    Execution harness: Python

  2. Task 2

    Input: {'clip_id': 'string', 'extracted_events': [{'type': 'string', 'timestamp': int, 'details': {}}]}

    Output: {'clip_id': 'string', 'strategic_summary': 'string', 'recommendations': [{'area': 'string', 'insight': 'string'}]}

    Execution harness: Python

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Agents for Dynamic Spatial Strategy Analysis