- **Bootstrap IoC-based architecture with modular services.**

- **Implement `MlxModelService` for local LLM backend.**
- **Introduce `DatabaseService` for MySQL integration.**
- **Add `HistoryService` to manage conversation context.**
- **Set up CLI interface via `CliUiService`.**
- **Establish EventBus for token streaming.**
- **Include conversation repository for data persistence.**
- **Add environment-based configuration management.**
- **Draft IoC architectural plan.**

services/chat/chat_service.py

@@ -0,0 +1,35 @@
+from services.model.base import AbstractModelService
+from services.chat.history_service import HistoryService
+from services.events.event_bus import EventBus
+
+
+class ChatService:
+    """대화 오케스트레이션 서비스."""
+
+    EVENT_TOKEN = "stream.token"
+    EVENT_END = "stream.end"
+
+    def __init__(
+        self,
+        model: AbstractModelService,
+        history: HistoryService,
+        event_bus: EventBus,
+        max_tokens: int,
+    ):
+        self._model = model
+        self._history = history
+        self._event_bus = event_bus
+        self._max_tokens = max_tokens
+
+    def respond(self, user_input: str) -> str:
+        self._history.add("user", user_input)
+        prompt = self._model.build_prompt(self._history.get())
+
+        response_text = ""
+        for token in self._model.stream(prompt, self._max_tokens):
+            self._event_bus.publish(self.EVENT_TOKEN, token)
+            response_text += token
+
+        self._event_bus.publish(self.EVENT_END)
+        self._history.add("assistant", response_text)
+        return response_text

services/chat/compact_service.py

@@ -0,0 +1,18 @@
+from services.model.base import AbstractModelService
+
+
+class CompactService:
+    """오래된 대화 턴을 LLM으로 요약하는 서비스."""
+
+    def __init__(self, model: AbstractModelService, max_tokens: int = 512):
+        self._model = model
+        self._max_tokens = max_tokens
+
+    def summarize(self, turns: list[dict]) -> str:
+        text = "\n".join(f"{t['role']}: {t['content']}" for t in turns)
+        prompt_history = [
+            {"role": "system", "content": "당신은 대화 요약 전문가입니다."},
+            {"role": "user", "content": f"다음 대화의 핵심 내용을 한국어로 간결하게 요약해주세요:\n\n{text}"},
+        ]
+        prompt = self._model.build_prompt(prompt_history)
+        return "".join(self._model.stream(prompt, self._max_tokens))

services/chat/history_service.py

@@ -0,0 +1,81 @@
+from __future__ import annotations
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from services.db.conversation_repository import ConversationRepository
+    from services.chat.compact_service import CompactService
+
+
+class HistoryService:
+    """대화 히스토리를 관리하는 서비스."""
+
+    def __init__(
+        self,
+        system_prompt: str,
+        max_turns: int,
+        compact_threshold: int,
+        repository: ConversationRepository | None = None,
+        compact_service: CompactService | None = None,
+    ):
+        self._system_prompt = system_prompt
+        self._max_turns = max_turns
+        self._compact_threshold = compact_threshold
+        self._repository = repository
+        self._compact_service = compact_service
+        self._summary: str | None = None
+        self._turns: list[dict] = []
+        self._conversation_id: int | None = None
+
+        if repository:
+            self._load_or_create()
+
+    # ── DB 초기화 ────────────────────────────────────────────────
+
+    def _load_or_create(self) -> None:
+        conv_id = self._repository.get_latest_conversation_id()
+        if conv_id:
+            summary_id, summary = self._repository.get_latest_summary(conv_id)
+            turns = self._repository.load_turns_after(
+                conv_id, summary_id, self._compact_threshold * 2
+            )
+            self._summary = summary
+            self._turns = turns
+            self._conversation_id = conv_id
+        else:
+            self._conversation_id = self._repository.create_conversation()
+
+    # ── 공개 인터페이스 ───────────────────────────────────────────
+
+    def add(self, role: str, content: str) -> None:
+        self._turns.append({"role": role, "content": content})
+        if self._repository and self._conversation_id:
+            self._repository.save_message(self._conversation_id, role, content)
+        if role == "assistant":
+            self._maybe_compact()
+
+    def get(self) -> list[dict]:
+        msgs = [{"role": "system", "content": self._system_prompt}]
+        if self._summary:
+            msgs.append({"role": "system", "content": f"[이전 대화 요약]\n{self._summary}"})
+        msgs.extend(self._turns)
+        return msgs
+
+    def reset(self, new_conversation_id: int) -> None:
+        self._summary = None
+        self._turns = []
+        self._conversation_id = new_conversation_id
+
+    # ── 내부 ─────────────────────────────────────────────────────
+
+    def _maybe_compact(self) -> None:
+        if not self._compact_service or len(self._turns) <= self._compact_threshold:
+            return
+
+        mid = len(self._turns) // 2
+        old_turns, self._turns = self._turns[:mid], self._turns[mid:]
+
+        print("\n[대화 내용을 압축하는 중...]\n", flush=True)
+        self._summary = self._compact_service.summarize(old_turns)
+
+        if self._repository and self._conversation_id:
+            self._repository.save_summary(self._conversation_id, self._summary)