minor fix

openenv.yaml

@@ -29,18 +29,21 @@ tasks:
     difficulty: easy
     description: Classify a single content item into one violation category or CLEAN
     grader: "server.graders:single_label_entry"
+    grader_import: "server.graders.single_label_entry"
 
   - id: multi-label-classify
     name: multi-label-classify
     difficulty: medium
     description: Assign all applicable violation labels to content that may violate multiple policies
     grader: "server.graders:multi_label_entry"
+    grader_import: "server.graders.multi_label_entry"
 
   - id: ad-policy-compliance
     name: ad-policy-compliance
     difficulty: medium_hard
     description: Review ad copy against ad policies, identify violations, and cite specific rule IDs
     grader: "server.graders:ad_policy_entry"
+    grader_import: "server.graders.ad_policy_entry"
 
   - id: thread-moderation-hard
     name: thread-moderation-hard
@@ -49,6 +52,7 @@ tasks:
       Moderate a full WhatsApp conversation thread with growing context window.
       Handles cultural nuance, multi-label violations, and conflicting policy resolution.
     grader: "server.graders:thread_hard_entry"
+    grader_import: "server.graders.thread_hard_entry"
 
 observation_space:
   type: object

requirements.txt

@@ -1,6 +1,7 @@
 # This file was autogenerated by uv via the following command:
 #    uv export --format requirements-txt --no-hashes -o requirements.txt
-# -e .
+# editable install of this package so external checkers can import `server`
+-e .
 aiofile==3.9.0
     # via py-key-value-aio
 aiofiles==24.1.0

scripts/validate_graders.py

@@ -0,0 +1,83 @@
+import importlib
+import sys
+from pathlib import Path
+import yaml
+
+ROOT = Path(__file__).parent.parent
+OPENENV = ROOT / "openenv.yaml"
+
+# Ensure repo root is on sys.path so imports like `server.graders` work when run from scripts/
+sys.path.insert(0, str(ROOT))
+
+if not OPENENV.exists():
+    print(f"openenv.yaml not found at {OPENENV}")
+    sys.exit(2)
+
+with OPENENV.open("r", encoding="utf-8") as f:
+    cfg = yaml.safe_load(f)
+
+tasks = cfg.get("tasks", [])
+if not tasks:
+    print("No tasks found in openenv.yaml")
+    sys.exit(2)
+
+print(f"Found {len(tasks)} tasks. Validating grader importability...")
+
+failed = []
+for t in tasks:
+    tid = t.get("id") or t.get("name") or "<unknown>"
+    grader_spec = t.get("grader")
+    grader_dot = t.get("grader_import")
+    print(f"\n- Task: {tid}")
+
+    tried = []
+    ok = False
+
+    for spec in (grader_spec, grader_dot):
+        if not spec:
+            continue
+        tried.append(spec)
+        # Support both "module:callable" and "module.callable"
+        module = None
+        attr = None
+        if ":" in spec:
+            module, attr = spec.split(":", 1)
+        elif "." in spec:
+            parts = spec.rsplit(".", 1)
+            if len(parts) == 2:
+                module, attr = parts
+            else:
+                module = spec
+                attr = None
+        else:
+            module = spec
+            attr = None
+
+        try:
+            m = importlib.import_module(module)
+            if attr:
+                if hasattr(m, attr):
+                    print(f"  OK: imported {module} and found attribute '{attr}'")
+                    ok = True
+                    break
+                else:
+                    print(f"  ERROR: imported {module} but attribute '{attr}' not found")
+            else:
+                print(f"  OK: imported module {module}")
+                ok = True
+                break
+        except Exception as e:
+            print(f"  ERROR importing {module}: {e}")
+
+    if not ok:
+        failed.append((tid, tried))
+
+print("\nSummary:")
+if not failed:
+    print("All graders importable ✅")
+    sys.exit(0)
+else:
+    print(f"{len(failed)} task(s) failed to import graders:")
+    for tid, tried in failed:
+        print(f" - {tid}: tried {tried}")
+    sys.exit(1)

server/app.py

@@ -27,6 +27,26 @@ from server.models import (
     StepResult,
 )
 
+# Try to import OpenEnv core package at runtime and log availability.
+# Prefer the new `openenv.core` import; fall back to legacy `openenv_core` if needed.
+_oe_imported = False
+for _mod in ("openenv.core", "openenv_core"):
+    try:
+        _openenv_core = __import__(_mod, fromlist=["*"])
+        _oe_ver = getattr(_openenv_core, "__version__", None)
+        if _oe_ver:
+            print(f"[startup] {_mod} available, version={_oe_ver}")
+        else:
+            print(f"[startup] {_mod} imported (version unknown)")
+        _oe_imported = True
+        break
+    except Exception as _e:
+        # continue to next candidate
+        _oe_last_exc = _e
+
+if not _oe_imported:
+    print(f"[startup] openenv package not importable: {_oe_last_exc}")
+
 # ─── App Init ─────────────────────────────────────────────────────────────────
 
 app = FastAPI(