ColonelBlotto

Description

ColonelBlotto is an ORS environment for evaluating agents on playing Colonel Blotto, a resource allocation strategy game, against an LLM opponent. This environment wraps the ColonelBlotto implementation from TextArena, a framework for text-based game environments.

Capabilities

• Strategic resource allocation across multiple objectives
• Game-theoretic reasoning about opponent strategy
• Optimization under constraints (fixed total troops)
• Competitive two-player gameplay against an LLM opponent

Compute Requirements

ColonelBlotto does not require a sandbox. It has minimal compute requirements.

License

MIT.

Tasks

There are two splits: train (600 tasks) and test (600 tasks). Each split contains 50 tasks across each of 12 variants:

• ColonelBlotto-v0-small
• ColonelBlotto-v0-small-train
• ColonelBlotto-v0-small-raw
• ColonelBlotto-v0
• ColonelBlotto-v0-train
• ColonelBlotto-v0-raw
• ColonelBlotto-v0-large
• ColonelBlotto-v0-large-train
• ColonelBlotto-v0-large-raw
• ColonelBlotto-v0-extreme
• ColonelBlotto-v0-extreme-train
• ColonelBlotto-v0-extreme-raw

Each task is seeded for reproducibility.

Reward Structure

This is a sparse reward environment. Rewards are mapped from TextArena's native range of {-1, 0, 1} to {0.0, 0.5, 1.0} via (raw + 1) / 2.

We do not use LLM graders for this environment; reward is determined programmatically.

Data

Game state is generated procedurally by the TextArena engine using seeded randomness. No external data files are required.

Tools

Agents are given a single tool:

• allocate_troops(allocation): Allocate troops across fields. Format: 'A4 B2 C2' — troops must sum to the required total.

Time Horizon

ColonelBlotto is a multi-turn environment.

Environment Difficulty

Medium

Other Environment Requirements

This environment requires an OpenAI API key (passed via secrets) to power the LLM opponent.

Safety

Agents in ColonelBlotto interact only with a strategy game and have no access to external systems, the internet, or sensitive data. The environment does not present safety risks.

Citations

@software{textarena2024,
  author    = {Guertler, Leon and Banting, Wilfried and Pignatelli, Eduardo},
  title     = {TextArena},
  year      = {2024},
  publisher = {GitHub},
  url       = {https://github.com/LeonGuertler/TextArena}
}