Initial commit: audio-chat with fixes

- Created AGENTS.md with architecture documentation
- Fixed race conditions and async patterns
- Added conversation history to LLM prompts
- Fixed TTS audio shape handling
- Added buffer limits and graceful shutdown
- Fixed client.py with file sending support
- Removed duplicate requirements
- Added .gitignore

audio/stream.py

@@ -0,0 +1,50 @@
+import threading
+import time
+from config import Config
+
+
+class AudioStreamBuffer:
+    def __init__(self):
+        self.config = Config()
+        self.buffer = b""
+        self.lock = threading.Lock()
+        self.event = threading.Event()
+        self.running = False
+        # Limit buffer to 10 seconds of audio to prevent OOM
+        self.max_buffer_bytes = int(self.config.SAMPLE_RATE * 10 * 2)
+
+    def start(self):
+        self.running = True
+        self.buffer = b""
+        self.event.clear()
+
+    def add_chunk(self, chunk: bytes):
+        with self.lock:
+            self.buffer += chunk
+            # Evict oldest data if buffer exceeds limit
+            if len(self.buffer) > self.max_buffer_bytes:
+                self.buffer = self.buffer[-self.max_buffer_bytes // 2:]
+            if len(self.buffer) >= self.config.CHUNK_SIZE:
+                self.event.set()
+
+    def get_ready_chunk(self, timeout: float = 1.0) -> bytes:
+        if self.event.wait(timeout=timeout):
+            with self.lock:
+                chunk = self.buffer
+                self.buffer = b""
+                self.event.clear()
+                return chunk
+        return b""
+
+    def get_full_buffer(self) -> bytes:
+        with self.lock:
+            chunk = self.buffer
+            self.buffer = b""
+            return chunk
+
+    def stop(self):
+        self.running = False
+        self.event.set()
+
+    def is_running(self):
+        return self.running