fix(quotas): keyless agent chat bucket, keep disabled policies disabled, cached rates

- check_usage treats a request through a keyless (draft) agent as agent
  traffic, matching how its usage rows are bucketed and the headless rule
- dashboard edits carry the stored enabled flag instead of re-enabling the
  policy; disabled policies are labelled in the Quotas tab and the editor
- quota 429s send x-should-retry: false so OpenAI SDK clients do not retry
  a refusal that cannot succeed before the reset
- cached-input and cache-write rates for Anthropic, OpenRouter and Groq
  gpt-oss-120b; refresh OpenRouter deepseek-v3.2 list prices
- UsageQuota reuses usagePercent; docs note that a user override needs an
  existing user

docsgpt/api/answer/routes/answer.py

@@ -132,7 +132,9 @@ class AnswerResource(Resource, BaseAnswerResource):
                     return make_response({"error": "Unauthorized"}, 401)
 
                 if error := self.check_usage(
-                    processor.agent_config, processor.decoded_token
+                    processor.agent_config,
+                    processor.decoded_token,
+                    agent_id=processor.agent_id,
                 ):
                     return error
 

docsgpt/api/answer/routes/base.py

@@ -115,7 +115,10 @@ class BaseAnswerResource:
         return prepared
 
     def check_usage(
-        self, agent_config: Dict, decoded_token: Optional[Dict] = None
+        self,
+        agent_config: Dict,
+        decoded_token: Optional[Dict] = None,
+        agent_id: Optional[str] = None,
     ) -> Optional[Response]:
         """Refuse the request when a usage limit is exhausted.
 
@@ -127,6 +130,8 @@ class BaseAnswerResource:
             agent_config: The config dict of agent instance
             decoded_token: The request's resolved identity; its ``sub`` is the
                 billable user.
+            agent_id: The agent the request runs through. A draft agent has no
+                key, but its usage rows carry the agent id, so it is agent traffic.
 
         Returns:
             None or Response if either of limits exceeded.
@@ -134,7 +139,7 @@ class BaseAnswerResource:
         """
         api_key = agent_config.get("user_api_key")
         user_id = (decoded_token or {}).get("sub") or agent_config.get("user_id")
-        exceeded = QuotaService.check(user_id, "agent" if api_key else "direct")
+        exceeded = QuotaService.check(user_id, "agent" if api_key or agent_id else "direct")
         if exceeded is not None:
             return quota_exceeded_response(exceeded)
         if not api_key:
@@ -227,7 +232,9 @@ class BaseAnswerResource:
         Returns:
             None, or the refusal Response.
         """
-        error = self.check_usage(processor.agent_config, processor.decoded_token)
+        error = self.check_usage(
+            processor.agent_config, processor.decoded_token, agent_id=processor.agent_id
+        )
         if error is None or not conversation_id:
             return error
         user = processor.initial_user_id or (processor.decoded_token or {}).get("sub")

docsgpt/api/answer/routes/stream.py

@@ -154,7 +154,9 @@ class StreamResource(Resource, BaseAnswerResource):
                 )
 
             if error := self.check_usage(
-                processor.agent_config, processor.decoded_token
+                processor.agent_config,
+                processor.decoded_token,
+                agent_id=processor.agent_id,
             ):
                 return error
             should_persist, visibility = resolve_persistence(

docsgpt/api/v1/routes.py

@@ -344,7 +344,9 @@ def chat_completions():
         if claimed_conversation_id:
             usage_error = helper.check_usage_on_resume(processor, claimed_conversation_id)
         else:
-            usage_error = helper.check_usage(processor.agent_config, processor.decoded_token)
+            usage_error = helper.check_usage(
+                processor.agent_config, processor.decoded_token, agent_id=processor.agent_id
+            )
         if usage_error:
             return usage_error