0

Slitherlink

Fresh

Slitherlink is an environment for evaluating agents on logic puzzle solving with loop-drawing constraints.

Type
RL Env
Runtime
TextArena
License
unknown
Size
300 tasks
Published
Mar 2026

Cite

Notes

Only stored in your browser.

Slitherlink

OpenReward Environment

Description

Slitherlink is an environment for evaluating agents on logic puzzle solving with loop-drawing constraints. This environment wraps the Slitherlink implementation from TextArena, a framework for text-based game environments.

Capabilities

  • Constraint-based logic puzzle solving
  • Understanding topological constraints (closed loops)
  • Spatial reasoning with edge relationships
  • Systematic constraint propagation

Compute Requirements

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

License

MIT.

Tasks

The number of variants is dynamically determined from the TextArena registry (all variants with "Slitherlink" prefix). There are two splits: train and test. Each split contains 50 tasks per variant.

Each task is seeded for reproducibility.

Reward Structure

This is a sparse reward environment. The agent receives a reward of 1.0 upon successfully completing the puzzle and 0.0 otherwise.

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:

  • toggle_edge(orientation, row, column): Toggle an edge on the Slitherlink grid.

Time Horizon

Slitherlink is a multi-turn environment.

Environment Difficulty

This environment presents moderate to challenging difficulty, requiring agents to satisfy local constraints (numbers in cells) while maintaining global constraints (forming a single closed loop).

Other Environment Requirements

There are no further environment requirements; Slitherlink works out of the box without any secrets or API keys.

Safety

Agents in Slitherlink interact only with a logic puzzle 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}
}