add default sandbox_labels to rlm-secrets

environments/rlm_secrets/README.md

@@ -93,3 +93,9 @@ This environment is specifically designed to test RLM capabilities:
 5. **Tests information flow**: Data must flow: file → sub-LLM → root-LLM → tool → answer
 
 The puzzle is simple enough that models should be able to solve it, while being complex enough to exercise all RLM components.
+
+## Changelog
+
+- v0.1.1 (01 Feb 2026):
+  - add default "rlm-secrets" label to the `sandbox_labels` no matter what the user passes ther in the kwargs
+  - dedupe `sandbox_labels` if passed via the kwargs

environments/rlm_secrets/pyproject.toml

@@ -2,7 +2,7 @@
 name = "rlm-secrets"
 description = "File puzzle environment for testing RLM capabilities: root tools, sub-LLM tools, file operations"
 tags = ["multi-turn", "rlm", "tools", "eval"]
-version = "0.1.0"
+version = "0.1.1"
 requires-python = ">=3.10"
 dependencies = [
     "verifiers>=0.1.8",

environments/rlm_secrets/rlm_secrets.py

@@ -483,6 +483,16 @@ def load_environment(
         weights=[0.5, 0.5],
     )
 
+    sandbox_labels = kwargs.pop("sandbox_labels", [])
+    if not (
+        isinstance(sandbox_labels, list)
+        and all(isinstance(label, str) for label in sandbox_labels)
+    ):
+        raise ValueError(
+            f"sandbox_labels must be of type list[str]; you provided {sandbox_labels}"
+        )
+    sandbox_labels = list(set(["rlm-secrets"] + sandbox_labels))
+
     return RLMSecretsEnv(
         dataset=train_dataset,
         num_files=num_files,
@@ -492,5 +502,6 @@ def load_environment(
         sub_tool_max_turns=sub_tool_max_turns,
         max_sub_llm_parallelism=max_sub_llm_parallelism,
         code_execution_timeout=code_execution_timeout,
+        sandbox_labels=sandbox_labels,
         **kwargs,
     )