diag: temp live-frames route (frame timings only, no key exposure)

node_probe.py

@@ -99,3 +99,50 @@ def register_routes(fa_app):
             )
         except Exception as e:
             return JSONResponse({"error": str(e)}, status_code=502)
+
+    @fa_app.get("/respite/v2/live-frames")
+    async def _live_frames(q: str = "Write three short paragraphs about rivers."):
+        """TEMP DIAGNOSTIC: open a Live socket, log outputTranscription frame
+        timings/count. Returns only timing metadata — never the API key."""
+        import asyncio, json, time
+        websockets = None
+        try:
+            import websockets as _ws
+            websockets = _ws
+        except ImportError:
+            return JSONResponse({"error": "websockets not installed"})
+        key = os.environ.get("GOOGLE_API_KEY") or os.environ.get("GEMINI_API_KEY")
+        if not key:
+            return JSONResponse({"error": "no google key in env"})
+        url = f"wss://generativelanguage.googleapis.com/ws/google.ai.generativelanguage.v1beta.GenerativeService.BidiGenerateContent?key={key}"
+        frames = []
+        t0 = time.monotonic()
+
+        async def run():
+            async with websockets.connect(url, max_size=None) as ws:
+                await ws.send(json.dumps({"setup": {"model": "models/gemini-3.8-live"}}))
+                setup = True
+                while True:
+                    try:
+                        raw = await asyncio.wait_for(ws.recv(), timeout=45)
+                    except (asyncio.TimeoutError, Exception):
+                        break
+                    try:
+                        f = json.loads(raw)
+                    except Exception:
+                        continue
+                    if f.get("setupComplete"):
+                        await ws.send(json.dumps({"clientContent": {"turns": [{"role": "user", "parts": [{"text": q}]}], "turnComplete": True}}))
+                        setup = False
+                        continue
+                    t = f.get("serverContent", {}).get("outputTranscription", {}).get("text")
+                    if t:
+                        frames.append({"t": round(time.monotonic() - t0, 2), "len": len(t)})
+                    if f.get("serverContent", {}).get("generationComplete"):
+                        break
+
+        try:
+            await asyncio.wait_for(run(), timeout=50)
+        except Exception as e:
+            return JSONResponse({"error": str(e), "frames": frames})
+        return JSONResponse({"frames": frames, "count": len(frames), "total_chars": sum(f["len"] for f in frames)})