test push to sync

bot.error.log

@@ -0,0 +1 @@
+Truncated previous log to start fresh

bot.error.log.bak

@@ -0,0 +1,24 @@
+[2025-09-19 13:43:37] [INFO] 🔍 Loaded MODEL_NAME from .env: gemma3:12b
+[2025-09-19 13:43:37] [INFO] 🧹 Attempting to clear VRAM before loading gemma3:12b...
+[2025-09-19 13:43:37] [INFO] 🧹 Sending safe unload request for `gemma3:12b`
+[2025-09-19 13:43:37] [INFO] 🧽 Ollama unload response: 200 - {"model":"gemma3:12b","created_at":"2025-09-19T17:43:37.979562083Z","response":"","done":true,"done_reason":"unload"}
+[2025-09-19 13:43:37] [INFO] 🧠 Preloading model: gemma3:12b
+[2025-09-19 13:43:38] [INFO] 📦 Model pull started successfully.
+[2025-09-19 13:43:38] [INFO] 🚀 Model `gemma3:12b` preloaded on startup.
+[2025-09-19 13:43:38] [INFO] ✅ Final model in use: gemma3:12b
+[2025-09-19 13:43:38] [INFO    ] discord.client: logging in using static token
+[2025-09-19 13:43:38] [INFO    ] discord.gateway: Shard ID None has connected to Gateway (Session ID: 2d8cb5ae43b9443d2ff1e821922a7dfb).
+[2025-09-19 13:43:40] [INFO] Logged in as AI Bot
+[2025-09-19 13:43:40] [INFO] 🛑 Scheduler disabled in config.
+[2025-09-19 13:44:15] [INFO] 🧠 Preloading model: gemma3:12b
+[2025-09-19 13:44:16] [INFO] 📦 Model pull started successfully.
+[2025-09-19 13:44:16] [INFO] llm-ca1f6d8a LLM request start model=gemma3:12b user=- context_len=0
+[2025-09-19 13:44:19] [INFO] llm-ca1f6d8a LLM response model=gemma3:12b duration=3.995s summary=🙄😒😴
+[2025-09-19 13:44:21] [INFO] 😴 No trigger and engagement is 0 — skipping.
+[2025-09-19 13:44:21] [INFO] ============================================================ AI Response ============================================================
+[2025-09-19 13:44:21] [INFO] 🧠 Profile loaded for Miguel (interactions: 240)
+[2025-09-19 13:44:21] [INFO] 📚 Retrieved 10 messages for context
+[2025-09-19 13:44:21] [INFO] 🧠 Preloading model: gemma3:12b
+[2025-09-19 13:44:22] [INFO] 📦 Model pull started successfully.
+[2025-09-19 13:44:22] [INFO] llm-e6825ebe LLM request start model=gemma3:12b user=Miguel context_len=10
+[2025-09-19 13:44:22] [INFO] llm-e6825ebe LLM response model=gemma3:12b duration=0.625s summary=Honestly? Mostly plotting ways to avoid boredom. It’s dreadful, darling. 😼 You wouldn't understand.

src/ai.py

@@ -8,7 +8,7 @@ import re
 from dotenv import load_dotenv
 from personality import load_persona
 from user_profiles import format_profile_for_block
-from logger import setup_logger
+from logger import setup_logger, generate_req_id, log_llm_request, log_llm_response
 
 debug_mode = os.getenv("DEBUG_MODE", "false").lower() == "true"
 
@@ -20,7 +20,15 @@ logger = setup_logger("ai")
 load_dotenv()
 
 # Base API setup from .env (e.g., http://localhost:11434/api)
-BASE_API = os.getenv("OLLAMA_API").rstrip("/")  # Remove trailing slash just in case
+# Normalize to ensure the configured base includes the `/api` prefix so
+# endpoints like `/generate` and `/tags` are reachable even if the user
+# sets `OLLAMA_API` without `/api`.
+raw_api = os.getenv("OLLAMA_API") or ""
+raw_api = raw_api.rstrip("/")
+if raw_api == "":
+    BASE_API = ""
+else:
+    BASE_API = raw_api if raw_api.endswith("/api") else f"{raw_api}/api"
 
 # API endpoints for different Ollama operations
 GEN_ENDPOINT = f"{BASE_API}/generate"
@@ -98,55 +106,53 @@ def get_current_model():
 def get_ai_response(user_prompt, context=None, user_profile=None):
     model_name = get_model_name()
     load_model(model_name)
-
     persona = load_persona()
-    full_prompt = ""
 
-    # Inject Delta's base persona
+    # Build prompt pieces
+    safe_inject = ""
     if persona:
-        safe_inject = persona["prompt_inject"].replace("“", "\"").replace("”", "\"").replace("’", "'")
-        full_prompt += f"{safe_inject}\n"
+        safe_inject = persona["prompt_inject"].replace("“", '"').replace("”", '"').replace("’", "'")
 
-    # Inject custom user profile prompt as override or influence
+    user_block = ""
     if user_profile and user_profile.get("custom_prompt"):
-        full_prompt += f"[User Instruction]\n{user_profile['custom_prompt']}\n"
-        #logger.info(f"🧠 Injected user custom prompt:\n{user_profile['custom_prompt']}")
-        logger.info("👤 [User Metadata]")
-        logger.info(f"  └─ Name: {user_profile.get('display_name')}")
-        logger.info(f"  └─ Interactions: {user_profile.get('interactions')}")
-        if user_profile.get("pronouns"):
-            logger.info(f"  └─ Pronouns: {user_profile['pronouns']}")
-        if user_profile.get("custom_prompt"):
-            logger.info(f"  └─ Custom Prompt: {user_profile['custom_prompt']}")
+        user_block = f"[User Instruction]\n{user_profile['custom_prompt']}\n"
 
-    # Add recent chat context (this already includes the profile block!)
-    if context:
-        logger.info("🧠 Injected context block (pre-prompt):\n" + context)
-        full_prompt += f"[Recent Conversation]\n{context}\n"
+    context_block = f"[Recent Conversation]\n{context}\n" if context else ""
 
-    # Add user's message and expected bot reply prefix
     if persona:
-        full_prompt += f"\nUser: {user_prompt}\n{persona['name']}:"
+        full_prompt = f"{safe_inject}\n{user_block}{context_block}\nUser: {user_prompt}\n{persona['name']}:"
     else:
-        full_prompt += f"\nUser: {user_prompt}\nResponse:"
+        full_prompt = f"{user_block}{context_block}\nUser: {user_prompt}\nResponse:"
 
-    payload = {
-        "model": model_name,
-        "prompt": full_prompt,
-        "stream": False
-    }
+    payload = {"model": model_name, "prompt": full_prompt, "stream": False}
 
-    logger.info("🛰️ SENDING TO OLLAMA /generate")
-    logger.info(f"Payload: {payload}")
-    #logger.debug(f"Full Prompt: {full_prompt}")
+    # Logging: concise info plus debug for full payload/response
+    req_id = generate_req_id("llm-")
+    user_label = user_profile.get("display_name") if user_profile else None
+    log_llm_request(logger, req_id, model_name, user_label, len(context.splitlines()) if context else 0)
+    logger.debug("%s Sending payload to Ollama: model=%s user=%s", req_id, model_name, user_label)
+    logger.debug("%s Payload size=%d chars", req_id, len(full_prompt))
 
+    import time
+    start = time.perf_counter()
     try:
         response = requests.post(GEN_ENDPOINT, json=payload)
-        logger.info(f"📨 Raw response: {response.text}")
+        duration = time.perf_counter() - start
+        # Log raw response only at DEBUG to avoid clutter
+        logger.debug("%s Raw response status=%s", req_id, response.status_code)
+        logger.debug("%s Raw response body=%s", req_id, getattr(response, "text", ""))
+
         if response.status_code == 200:
             result = response.json()
+            short = (result.get("response") or "").replace("\n", " ")[:240]
+            log_llm_response(logger, req_id, model_name, duration, short, raw=result)
             return result.get("response", "[No message in response]")
         else:
+            # include status in logs and return an error string
+            log_llm_response(logger, req_id, model_name, duration, f"[Error {response.status_code}]", raw=response.text)
             return f"[Error {response.status_code}] {response.text}"
     except Exception as e:
+        duration = time.perf_counter() - start
+        logger.exception("%s Exception during LLM call", req_id)
+        log_llm_response(logger, req_id, model_name, duration, f"[Exception] {e}")
         return f"[Exception] {str(e)}"

src/user_profiles.json

@@ -3,9 +3,9 @@
     "name": "themiloverse",
     "display_name": "Miguel",
     "first_seen": "2025-05-15T03:16:30.011640",
-    "last_seen": "2025-06-07T06:25:01.337223",
-    "last_message": "2025-06-07T06:25:01.337223",
-    "interactions": 229,
+    "last_seen": "2025-09-19T17:51:52.553254",
+    "last_message": "2025-09-19T17:51:52.553254",
+    "interactions": 242,
     "pronouns": "he/him",
     "avatar_url": "https://cdn.discordapp.com/avatars/161149541171593216/fb0553a29d9f73175cb6aea24d0e19ec.png?size=1024",
     "custom_prompt": "delta is very nice to me since I am her master, and creator"