env_id: code-debug-env
version: "1.0.0"
description: "RL environment where AI agents debug Python code"

tasks:
  - id: fix_addition_bug
    difficulty: easy
    max_steps: 5
    description: "Fix the function so it correctly returns the sum of two numbers"
    grader:
      type: programmatic
      prompt_template: "Score the fix 0.0 to 1.0 based on correctness"

  - id: fix_max_finder_bug
    difficulty: medium
    max_steps: 5
    description: "Fix the function so it correctly returns the largest number in a list"
    grader:
      type: programmatic
      prompt_template: "Score the fix 0.0 to 1.0 based on correctness"

  - id: fix_fibonacci_bug
    difficulty: hard
    max_steps: 5
    description: "Fix the recursive fibonacci function"
    grader:
      type: programmatic
      prompt_template: "Score the fix 0.0 to 1.0 based on correctness"