feat: store and display backend + host metadata on chat messages

Each assistant message in the session JSON now carries:
  backend, backend_label, host (platform.node())

These fields are shown as model tags in the UI — on live responses and
when loading session history. Session log entries (sessions/YYYY-MM-DD.md)
include the backend label and host in the turn header.

The local (OpenAI-compat) backend strips non-standard fields before
sending messages to the API so extra fields don't leak upstream.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

cortex/llm_client.py

@@ -183,7 +183,8 @@ async def _local(system_prompt: str, messages: list[dict], model_cfg: dict | Non
     msgs: list[dict] = []
     if system_prompt:
         msgs.append({"role": "system", "content": system_prompt})
-    msgs.extend(messages)
+    # Strip any non-standard metadata fields before sending to the API
+    msgs.extend({"role": m["role"], "content": m["content"]} for m in messages)
 
     url = api_url.rstrip("/") + chat_path
     headers: dict[str, str] = {}

cortex/routers/chat.py

@@ -1,5 +1,6 @@
 import asyncio
 import json
+import platform
 import jwt
 from fastapi import APIRouter, HTTPException, Query, Request
 from fastapi.responses import StreamingResponse
@@ -108,10 +109,18 @@ async def _stream_chat(req: ChatRequest):
 
         try:
             response_text, actual_backend = task.result()
-            history.append({"role": "assistant", "content": response_text})
+            backend_label = _backend_label(actual_backend, user, role="chat")
+            host = platform.node()
+            history.append({
+                "role": "assistant",
+                "content": response_text,
+                "backend": actual_backend,
+                "backend_label": backend_label,
+                "host": host,
+            })
             save_session(session_id, history)
             if not req.off_record:
-                log_turn(session_id, req.message, response_text)
+                log_turn(session_id, req.message, response_text, backend_label, host)
 
             # fallback_used only makes sense for explicit backend selections.
             # In auto mode (req.model is None), just report what responded.
@@ -121,7 +130,8 @@ async def _stream_chat(req: ChatRequest):
                 "response": response_text,
                 "session_id": session_id,
                 "backend": actual_backend,
-                "backend_label": _backend_label(actual_backend, user, role="chat"),
+                "backend_label": backend_label,
+                "host": host,
                 "fallback_used": fallback_used,
             }
             yield f"data: {json.dumps(payload)}\n\n"

cortex/session_logger.py

@@ -3,7 +3,13 @@ from config import settings
 from persona import persona_path
 
 
-def log_turn(session_id: str, user_msg: str, assistant_msg: str) -> None:
+def log_turn(
+    session_id: str,
+    user_msg: str,
+    assistant_msg: str,
+    backend_label: str = "",
+    host: str = "",
+) -> None:
     today = datetime.now().strftime("%Y-%m-%d")
     sessions_dir = persona_path() / "sessions"
     sessions_dir.mkdir(exist_ok=True)
@@ -12,11 +18,14 @@ def log_turn(session_id: str, user_msg: str, assistant_msg: str) -> None:
     timestamp = datetime.now().strftime("%H:%M")
     is_new = not log_file.exists()
 
+    meta_parts = [p for p in [backend_label, host] if p]
+    meta = f" · {' / '.join(meta_parts)}" if meta_parts else ""
+
     with open(log_file, "a") as f:
         if is_new:
             f.write(f"# Session Log — {today}\n")
         f.write(
-            f"\n### [{timestamp}] `{session_id}`\n"
+            f"\n### [{timestamp}] `{session_id}`{meta}\n"
             f"**{settings.user_name}:** {user_msg}\n\n"
             f"**{settings.agent_name}:** {assistant_msg}\n"
         )

cortex/static/app.js

@@ -566,6 +566,13 @@
                 currentHistory.push({ role, content: msg.content });
                 const msgDiv = addMessage(role, msg.content);
                 attachHistoryControls(msgDiv, i);
+                if (role === 'assistant' && (msg.backend_label || msg.backend)) {
+                    const modelTag = document.createElement('div');
+                    modelTag.className = 'model-tag';
+                    const label = msg.backend_label || msg.backend;
+                    modelTag.textContent = msg.host ? `${label} · ${msg.host}` : label;
+                    msgDiv.appendChild(modelTag);
+                }
             }
 
             if (!silent) addMessage('system', `Resumed session ${id}`);
@@ -979,9 +986,10 @@
                             const modelTag = document.createElement('div');
                             modelTag.className = 'model-tag' + (data.fallback_used ? ' fallback' : '');
                             const label = data.backend_label || data.backend || '';
+                            const hostSuffix = data.host ? ` · ${data.host}` : '';
                             modelTag.textContent = data.fallback_used
-                                ? `⚡ fallback → ${label}`
+                                ? `⚡ fallback → ${label}${hostSuffix}`
-                                : label;
+                                : `${label}${hostSuffix}`;
                             thinkingDiv.appendChild(modelTag);
                         } else if (data.type === 'error') {
                             throw new Error(data.message);