Implementation:Farama Foundation Gymnasium Check Env

Knowledge Sources	Gymnasium Gymnasium Utils
Domains	Reinforcement_Learning, Testing
Last Updated	2026-02-15 03:00 GMT

Overview

Concrete tool for validating custom RL environment implementations against the Gymnasium API contract.

Description

The check_env function performs comprehensive validation of a Gymnasium environment. It checks observation and action spaces, reset return types, seed determinism, step determinism, render mode compliance, and close safety. It raises AssertionError or TypeError with descriptive messages when violations are found and logs warnings for non-critical issues.

Usage

Call check_env after instantiating a custom environment (without make() wrappers). Use skip_render_check=True in environments without rendering support or in CI environments without display access.

Code Reference

Source Location

Repository: Gymnasium
File: gymnasium/utils/env_checker.py
Lines: L351-452

Signature

def check_env(
    env: gym.Env,
    warn: bool = None,
    skip_render_check: bool = False,
    skip_close_check: bool = False,
):
    """Check that an environment follows Gymnasium's API.

    Args:
        env: The environment instance to validate
        warn: Ignored (previously silenced warnings)
        skip_render_check: Whether to skip render method checks
        skip_close_check: Whether to skip close method checks

    Raises:
        AssertionError: If environment violates the API contract
        TypeError: If environment doesn't inherit from gymnasium.Env
    """

Import

from gymnasium.utils.env_checker import check_env

I/O Contract

Inputs

Name	Type	Required	Description
env	gym.Env	Yes	Environment instance to validate (unwrapped preferred)
skip_render_check	bool	No	Skip render validation (default False)
skip_close_check	bool	No	Skip close validation (default False)

Outputs

Name	Type	Description
(none)	None	Passes silently if valid; raises exceptions on failure

Usage Examples

Validate Custom Environment

from gymnasium.utils.env_checker import check_env

# Create environment directly (not via make())
env = GridWorldEnv(size=5, render_mode="rgb_array")

# Run full validation
check_env(env)

# Skip render check for headless CI
check_env(env, skip_render_check=True)

Integration Test Pattern

import pytest
from gymnasium.utils.env_checker import check_env

def test_custom_env_api_compliance():
    env = MyCustomEnv()
    check_env(env, skip_render_check=True)
    env.close()

Related Pages

Implements Principle

Principle:Farama_Foundation_Gymnasium_Environment_Validation

Requires Environment

Environment:Farama_Foundation_Gymnasium_Python_3_10_Runtime

Uses Heuristic

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment