fix llm api call

docker-compose.registry.yml

@@ -1,8 +1,6 @@
-version: '3.8'
-
 services:
   mcp-server:
-    image: ghcr.io/${GITHUB_REPOSITORY:-your-username/sleeper-bot}-mcp-server:latest
+    image: ghcr.io/wombat45/sleeper-bot-mcp-server:latest
     container_name: sleeper-bot-mcp-server
     ports:
       - "8000:8000"
@@ -25,7 +23,7 @@ services:
     restart: unless-stopped
 
   llm-agent:
-    image: ghcr.io/${GITHUB_REPOSITORY:-your-username/sleeper-bot}-llm-agent:latest
+    image: ghcr.io/wombat45/sleeper-bot-llm-agent:latest
     container_name: sleeper-bot-llm-agent
     ports:
       - "8001:8000"
@@ -35,9 +33,9 @@ services:
       - API_KEY=${API_KEY}
       - MCP_SERVER_URL=http://mcp-server:8000
       - DEFAULT_LEAGUE_ID=${DEFAULT_LEAGUE_ID}
-      - LLM_PROVIDER=${LLM_PROVIDER:-ollama}
-      - LLM_URL=${LLM_URL:-http://host.docker.internal:11434/api/generate}
-      - LLM_API_KEY=${LLM_API_KEY:-}
+      - LLM_PROVIDER=${LLM_PROVIDER:-digitalocean}
+      - LLM_URL=${LLM_URL:-https://td4bxjc626e5xweih26eppev.agents.do-ai.run/api/v1/chat/completions}
+      - LLM_API_KEY=${LLM_API_KEY:-O1AD1a0JeIl3B-T9sJvvBgHEHfwIoPze}
       - LLM_MODEL=${LLM_MODEL:-llama3.2:latest}
     volumes:
       - ./logs:/app/logs
@@ -55,7 +53,7 @@ services:
     restart: unless-stopped
 
   discord-bot:
-    image: ghcr.io/${GITHUB_REPOSITORY:-your-username/sleeper-bot}-discord-bot:latest
+    image: ghcr.io/wombat45/sleeper-bot-discord-bot:latest
     container_name: sleeper-bot-discord-bot
     ports:
       - "8002:8000"

docker-compose.yml

@@ -43,7 +43,10 @@ services:
       - API_KEY=${API_KEY:-your-api-key-here}
       - MCP_SERVER_URL=http://mcp-server:8000
       - DEFAULT_LEAGUE_ID=${DEFAULT_LEAGUE_ID:-}
-      - LLM_URL=http://host.docker.internal:11434/api/generate
+      - LLM_PROVIDER=digitalocean
+      - LLM_URL=https://td4bxjc626e5xweih26eppev.agents.do-ai.run/api/v1/chat/completions
+      - LLM_API_KEY=O1AD1a0JeIl3B-T9sJvvBgHEHfwIoPze
+      - LLM_MODEL=llama3.2:latest
       - HOST=0.0.0.0
       - PORT=8000
     volumes:

llm-agent/agent.py

@@ -323,26 +323,66 @@ async def _generate_llm_response(self, query: str, data: Dict[str, Any]) -> str:
             print(f"🤖 LLM Prompt length: {len(prompt)} characters")
             print(f"🤖 LLM Prompt preview: {prompt[:200]}...")
 
-            # Call the LLM
+            # Get LLM configuration from environment
+            llm_provider = os.getenv("LLM_PROVIDER", "ollama")
+            llm_api_key = os.getenv("LLM_API_KEY", "")
+            llm_model = os.getenv("LLM_MODEL", "llama3.2:latest")
+
+            # Call the LLM based on provider
             async with httpx.AsyncClient() as client:
-                response = await client.post(
-                    LLM_URL,
-                    json={
-                        "model": "llama3.2:latest",  # Use the model you have installed
+                if llm_provider == "digitalocean":
+                    # DigitalOcean AI Agent format
+                    headers = {
+                        "Content-Type": "application/json",
+                        "Authorization": f"Bearer {llm_api_key}"
+                    }
+                    payload = {
+                        "messages": [
+                            {
+                                "role": "user",
+                                "content": prompt
+                            }
+                        ],
+                        "stream": False,
+                        "include_functions_info": False,
+                        "include_retrieval_info": False,
+                        "include_guardrails_info": False
+                    }
+                else:
+                    # Ollama format (default)
+                    headers = {"Content-Type": "application/json"}
+                    payload = {
+                        "model": llm_model,
                         "prompt": prompt,
                         "stream": False,
                         "options": {
                             "temperature": 0.7,
                             "top_p": 0.9,
-                            "max_tokens": 1000  # Increased from 500
+                            "max_tokens": 1000
                         }
-                    },
+                    }
+
+                response = await client.post(
+                    LLM_URL,
+                    json=payload,
+                    headers=headers,
                     timeout=30
                 )
 
                 if response.status_code == 200:
                     result = response.json()
-                    llm_response = result.get("response", "").strip()
+
+                    # Parse response based on provider
+                    if llm_provider == "digitalocean":
+                        # DigitalOcean format: response is in choices[0].message.content
+                        if "choices" in result and len(result["choices"]) > 0:
+                            llm_response = result["choices"][0]["message"]["content"].strip()
+                        else:
+                            llm_response = result.get("response", "").strip()
+                    else:
+                        # Ollama format
+                        llm_response = result.get("response", "").strip()
+
                     print(f"🤖 LLM Response length: {len(llm_response)} characters")
                     print(f"🤖 LLM Response preview: {llm_response[:200]}...")