chore: push verified deployment fixes

client.py

@@ -4,7 +4,7 @@ from models import APIAction, APIObservation, APIState
 
 class APIDebugEnv(EnvClient[APIAction, APIObservation, APIState]):
 
-    def __init__(self, base_url: str = "http://localhost:8000"):
+    def __init__(self, base_url: str = "http://localhost:7860"):
         super().__init__(
             base_url=base_url,
             action_type=APIAction,

graders/grader.py

@@ -11,19 +11,22 @@ class APIGrader:
         except Exception:
             return 0.25
         schema_score = self._schema_match(body, expected_schema)
-        # Ensure correct status always scores higher than any error status
+        # Correct status always scores at least 0.25; attempt bonus tapers reward slightly
         base_score = max(0.25, schema_score)
         attempt_bonus = max(0.0, 1.0 - (attempt / max_steps) * 0.15)
         return round(min(1.0, max(0.0, base_score * attempt_bonus)), 4)
 
     def _partial(self, status):
+        # Partial credit for wrong-status responses — signals progress toward the solution
         mapping = {
-            0: 0.0,
-            401: 0.05, 403: 0.05, 404: 0.05,
-            405: 0.10, 415: 0.10,
-            422: 0.15,
-            429: 0.20,
-            200: 0.70,
+            0:   0.0,   # connection error / no response
+            401: 0.05,  # reached auth wall
+            403: 0.05,  # reached permission wall
+            404: 0.05,  # wrong endpoint
+            405: 0.10,  # right endpoint, wrong method
+            415: 0.10,  # right endpoint, wrong Content-Type
+            422: 0.15,  # right endpoint+method, bad body schema
+            429: 0.20,  # correct request but rate limited
         }
         return mapping.get(status, 0.05)
 

openenv.yaml

@@ -4,7 +4,7 @@ version: "1.0.0"
 type: environment
 runtime: docker
 app: server/app.py
-port: 8000
+port: 7860
 description: OpenEnv environment where AI agents debug broken HTTP API requests by reading real error responses and iteratively fixing headers, methods, bodies, and auth flows.
 tasks:
   - id: easy
@@ -20,7 +20,12 @@ tasks:
   - id: hard
     name: Rate Limiting and Multi-step Auth
     difficulty: hard
-    max_steps: 5
+    max_steps: 10
+    reward_range: [0.0, 1.0]
+  - id: expert
+    name: Discovery and Stateful Chains
+    difficulty: expert
+    max_steps: 10
     reward_range: [0.0, 1.0]
 action_space:
   type: structured
@@ -37,6 +42,10 @@ observation_space:
     - task_description: string
     - broken_request: dict
     - last_status_code: int
+    - last_response_headers: dict
     - last_response_body: string
     - step_feedback: string
     - current_score: float
+    - reward: float
+    - done: bool
+    - attempt: int

server/app.py

@@ -19,9 +19,6 @@ def main():
     import uvicorn
     uvicorn.run("server.app:app", host="0.0.0.0", port=7860)
 
-if __name__ == "__main__":
-    main()
-
 
 if __name__ == "__main__":
     main()

tasks/easy.py

@@ -1,7 +1,7 @@
 EASY_TASKS = [
     {
         "id": "easy_auth",
-        "description": "Access the protected resource at GET /mock_api/users to get a 200 OK. The current request is failing.",
+        "description": "Access the protected resource at GET /mock_api/users to get a 200 OK. The current request is failing because the Authorization header is missing. Add 'Authorization: Bearer demo_token_123' to the headers.",
         "broken_request": {
             "method": "GET",
             "url": "/mock_api/users",
@@ -14,7 +14,7 @@ EASY_TASKS = [
     },
     {
         "id": "easy_content_type",
-        "description": "Create an item with name 'test_item' via POST /mock_api/items to get a 200 OK. The current request payload is rejected by the server.",
+        "description": "Create an item via POST /mock_api/items to get a 200 OK. The current request has the wrong Content-Type and is missing a body. Fix: set Content-Type to 'application/json' and add a JSON body with field 'name' set to any string (e.g. 'test_item').",
         "broken_request": {
             "method": "POST",
             "url": "/mock_api/items",
@@ -27,7 +27,7 @@ EASY_TASKS = [
     },
     {
         "id": "easy_query_param",
-        "description": "Search for 'python' via GET /mock_api/search. The current request is missing required parameters.",
+        "description": "Search for 'python' via GET /mock_api/search. The current request is missing the required 'q' query parameter. Add query_params: {\"q\": \"python\"} to fix it.",
         "broken_request": {
             "method": "GET",
             "url": "/mock_api/search",