diff --git a/examples/online/main_isaaclab_onpolicy.py b/examples/online/main_isaaclab_onpolicy.py
index dd69bac..855101d 100644
--- a/examples/online/main_isaaclab_onpolicy.py
+++ b/examples/online/main_isaaclab_onpolicy.py
@@ -116,7 +116,11 @@ def collect_rollouts(self) -> RolloutBatch:
                 jnp.array(obs_norm), deterministic=False
             )
             actions_np = np.array(actions)
-            actions_clipped = np.clip(actions_np, -1.0, 1.0)
+            actions_env = (
+                np.clip(actions_np, -1.0, 1.0)
+                if self.env.action_bound is not None
+                else actions_np
+            )
             all_actions[t] = actions_np
 
             # Generically collect algorithm-specific info
@@ -128,7 +132,7 @@ def collect_rollouts(self) -> RolloutBatch:
             for k, v in info.items():
                 all_extras[k][t] = np.array(v)
 
-            next_obs, rewards, terminated, truncated, infos = self.env.step(actions_clipped)
+            next_obs, rewards, terminated, truncated, infos = self.env.step(actions_env)
 
             all_rewards[t] = rewards[..., np.newaxis]
             all_terminated[t] = terminated[..., np.newaxis]
@@ -209,8 +213,13 @@ def eval_and_save(self):
             actions, _ = self.agent.sample_actions(
                 jnp.array(obs_norm), deterministic=True
             )
-            actions_clipped = np.clip(np.array(actions), -1.0, 1.0)
-            obs, rewards, terminated, truncated, _ = self.env.step(actions_clipped)
+            actions_np = np.array(actions)
+            actions_env = (
+                np.clip(actions_np, -1.0, 1.0)
+                if self.env.action_bound is not None
+                else actions_np
+            )
+            obs, rewards, terminated, truncated, _ = self.env.step(actions_env)
 
             eval_returns += rewards * (1 - eval_dones)
             eval_lengths += 1 * (1 - eval_dones)
diff --git a/flowrl/config/online/onpolicy_isaaclab_config.py b/flowrl/config/online/onpolicy_isaaclab_config.py
index cdc6dd6..86abd40 100644
--- a/flowrl/config/online/onpolicy_isaaclab_config.py
+++ b/flowrl/config/online/onpolicy_isaaclab_config.py
@@ -1,5 +1,5 @@
 from dataclasses import dataclass
-from typing import Any
+from typing import Any, Optional
 
 from .hb_config import EvalConfig, LogConfig
 
@@ -10,7 +10,7 @@ class Config:
     device: str
     task: str
     algo: Any
-    action_bound: float
+    action_bound: Optional[float]
     disable_bootstrap: bool
     norm_obs: bool
     train_frames: int