Add task_id/seed to action for stateless step

README.md

@@ -65,7 +65,7 @@ See `training/README.md` and `training/grpo_train.py` for a minimal GRPO run.
 
 ## Environment design
 
-- Action: a single pytest file plus a finalize flag.
+- Action: a single pytest file plus a finalize flag (optionally include `task_id` and `seed` for stateless HTTP calls).
 - Observation: task prompt, kill counts, reference pass/fail, runtime, and
   surviving mutant ids.
 - Reward: mutant kill rate minus penalties for false positives, runtime, and

mutationgym_env/models.py

@@ -10,6 +10,12 @@ class MutationGymAction(Action):
     """Agent action containing pytest code and a finalize flag."""
 
     tests_py: str = Field(..., description="Pytest tests as a single Python file.")
+    task_id: Optional[str] = Field(
+        default=None, description="Optional task id for stateless step calls."
+    )
+    seed: Optional[int] = Field(
+        default=None, description="Optional seed for implicit reset."
+    )
     finalize: bool = Field(
         default=True, description="If true, end the episode after scoring."
     )

mutationgym_env/server/mutationgym_environment.py

@@ -84,7 +84,7 @@ class MutationGymEnvironment(Environment):
         if self._current_task is None:
             # In stateless HTTP calls, the server may instantiate a fresh
             # environment per request. Fall back to an implicit reset.
-            self.reset()
+            self.reset(seed=action.seed, task_id=action.task_id)
 
         try:
             self._state = State(