{"data":{"kind":"file","path":"README.md","version_id":"gq8ihbsr766alfccdgfkekav","entry":{"name":"README.md","path":"README.md","is_directory":false,"size":1235,"modified_at":"2026-08-04T13:07:34.453000","content_hash":"65a86e16850accf563a8332db828ab97e0f594b113896629f908b06a8fe83fbe"},"entries":[],"content":"# Snake Game Agent\n\nA verifiers environment that evaluates an LLM's ability to play the classic Snake game by generating optimal move sequences.\n\n## Overview\n\nThe agent is given a textual representation of a Snake game board and must output a sequence of moves (UP, DOWN, LEFT, RIGHT) to:\n1. Navigate toward food\n2. Avoid walls and the snake's own body\n3. Maximize survival time and score\n\n## Task Format\n\nEach task presents:\n- A grid representation of the game board\n- Snake head position and body segments\n- Food location(s)\n- Grid dimensions\n\nThe agent must produce a valid move sequence as a JSON array.\n\n## Scoring\n\n- **Path Validity** (0.3): Are all moves valid (no wall/self collisions)?\n- **Food Collection** (0.4): How many food items were collected?\n- **Survival** (0.2): How many moves survived without collision?\n- **Efficiency** (0.1): Ratio of useful moves to total moves?\n\n## Levels\n\n- **Level 0**: 5x5 grid, no obstacles, single food\n- **Level 1**: 8x8 grid, simple obstacles, single food\n- **Level 2**: 10x10 grid, complex obstacles, multiple food items\n- **Level 3**: 12x12 grid, moving obstacles, timed challenges\n\n## Usage\n\n```python\nimport verifiers as vf\n\nenv = vf.load_environment(\"champa/snake-game-agent\")\n```\n","encoding":"utf-8","truncated":false,"total_bytes":1235},"status":null}