feat: OpenRouter engine, with OPENROUTER_API_KEY env var

backend/src/agents/character.py

@@ -3,7 +3,7 @@ import json
 import inspect
 from string import Template
 
-from inference import LlamaMessage, replicate_llama_completion
+from inference import LlamaMessage, replicate_completion
 from agent import Agent, AgentField
 
 # Info and defaults
@@ -137,7 +137,7 @@ class CharacterAgent(Agent):
             LlamaMessage("system", self.system_prompt),
             LlamaMessage("user", content)
         ]
-        response = await replicate_llama_completion(self.app, messages)
+        response = await replicate_completion(self.app, messages)
         return response
 
     async def received_message(self, message):
@@ -201,7 +201,7 @@ class CharacterAgent(Agent):
         # TODO: Inject system prompt before the last user message
         # If not any user message, combine with first system prompt
 
-        controller_response = await replicate_llama_completion(self.app, llama_messages)
+        controller_response = await replicate_completion(self.app, llama_messages)
 
         if self.debug:
             await self.app.send_chat_message(
@@ -282,7 +282,7 @@ class CharacterAgent(Agent):
         memories = [it.document for it in query_result]
 
         print("rerank memories")
-        rerank_response = await replicate_llama_completion(self.app, [
+        rerank_response = await replicate_completion(self.app, [
             LlamaMessage("system", rerank_system_prompt),
             LlamaMessage("user", rerank_prompt_template.substitute(
                 conversation=chat_history,