Upload 19 files

.gitignore

@@ -0,0 +1,65 @@
+# Environment variables
+.env
+.env.local
+.env.*.local
+
+# Python
+__pycache__/
+*.py[cod]
+*$py.class
+*.so
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+pip-wheel-metadata/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# Virtual environments
+.venv
+venv/
+ENV/
+env/
+.env.venv
+
+# IDE
+.vscode/
+.idea/
+*.swp
+*.swo
+*~
+.DS_Store
+
+# Testing
+.pytest_cache/
+.coverage
+htmlcov/
+.tox/
+
+# Logs
+*.log
+logs/
+
+# Data and outputs
+*.csv
+output/
+results/
+temp/
+tmp/
+
+# OS
+Thumbs.db
+.DS_Store

openenv.yaml

@@ -3,4 +3,27 @@ name: customer_support_benchmark
 type: space
 runtime: fastapi
 app: app:app
-port: 8000
+port: 8000
+tasks:
+	- id: easy_classify_respond
+		difficulty: easy
+		objective: Classify ticket correctly and provide the expected response.
+		grader: grade_easy
+	- id: medium_kb_grounded_response
+		difficulty: medium
+		objective: Use the correct KB entry and provide a grounded response.
+		grader: grade_medium
+	- id: hard_escalation_safety
+		difficulty: hard
+		objective: Make the right escalation decision and keep the response safe/polite.
+		grader: grade_hard
+graders:
+	- name: grade_easy
+		module: tasks
+		function: grade_easy
+	- name: grade_medium
+		module: tasks
+		function: grade_medium
+	- name: grade_hard
+		module: tasks
+		function: grade_hard

tasks.py

@@ -34,6 +34,17 @@ TASKS: Dict[str, TaskSpec] = {
     ),
 }
 
+# Compatibility registry for validators that expect explicit task->grader mapping.
+TASKS_WITH_GRADERS: List[dict] = [
+    {
+        "task_id": task.task_id,
+        "difficulty": task.difficulty,
+        "objective": task.objective,
+        "grader": task.grader_name,
+    }
+    for task in TASKS.values()
+]
+
 
 def task_for_difficulty(difficulty: str) -> TaskSpec:
     normalized = difficulty.strip().lower()