feat(agent): LLM-powered collection node with welcome stage and session-scoped delete (assignment 4)

- prompts/collection.md: versioned collection system prompt extracted
  from workflow node yQaBxfrWjbRr with source/version/porting notes
- llm.py: LLMService protocol; FakeLLMService (scripted, records
  calls); OpenAILLMService (async, wait_for timeout with cancellation,
  retry only on timeout/429/connection, errors normalized)
- collect.py: run_collect reference impl - one LLM call produces
  state prefix + body, controlled repair (max 1 retry) for missing
  prefix or illegal stage proposal, whitelist guard, CollectReplyError
  without raw model content; dialogue history + confirmed fields
  injected into prompt so the model skips known info (作业4-5)
- fields.py / clock.py: three-state field candidates (unknown/
  candidate/confirmed), per-type validation (count not inferred from
  description, future time rejected), injectable Clock
- graph.py: welcome node - first message returns fixed 1001 greeting
  without LLM (mirrors workflow is_first_message -> Assigned Reply#3),
  route empty stage to welcome
- endpoints/delete_session: backend switch - langgraph clears graph
  checkpoint, fastgpt path unchanged
- config: AGENT_LLM_PROVIDER fake|openai with validation
- untrack .env (secrets), move template to root .env.example

Verified:
- pytest 103 passed; smoke turns 1/2/1
- Real LLM e2e: POST /chat multi-turn context works
- Golden eval collection_skip_known_info: 12/12 cases pass
- Golden eval no_false: 8/10 cases pass; 4 confirm/photo cases fail
  at final step - model routes on short reply '对' (prompt-adherence
  gap, photo/confirm stages are assignment 5 scope); recorded in
  evals/results/20261020-144856

Known limits: field extraction from model reply not wired (stub);
photo/confirm nodes still scaffold; clarify counting is approximate.

src/core/config.py

@@ -18,6 +18,15 @@ class Config:
     # Chat backend switch（作业 2）
     # "fastgpt" 保持现有行为；"langgraph" 使用内存图 fake 回复。
     AGENT_BACKEND = os.getenv("AGENT_BACKEND", "fastgpt")
+
+    # LLM provider（作业 4）：fake 为脚本化假模型，离线可运行；
+    # openai 为真实 provider（OpenAI 兼容接口）。
+    AGENT_LLM_PROVIDER = os.getenv("AGENT_LLM_PROVIDER", "fake")
+    LLM_MODEL = os.getenv("LLM_MODEL", "fake-reply-model")
+    LLM_BASE_URL = os.getenv("LLM_BASE_URL", "")
+    LLM_API_KEY = os.getenv("LLM_API_KEY", "")
+    LLM_TIMEOUT_SECONDS = float(os.getenv("LLM_TIMEOUT_SECONDS", "30"))
+    LLM_MAX_RETRIES = int(os.getenv("LLM_MAX_RETRIES", "2"))
     
     @classmethod
     def validate(cls):
@@ -28,6 +37,17 @@ class Config:
             raise ValueError(
                 f"AGENT_BACKEND must be 'fastgpt' or 'langgraph', got {cls.AGENT_BACKEND!r}"
             )
+        if cls.AGENT_LLM_PROVIDER not in ("fake", "openai"):
+            raise ValueError(
+                "AGENT_LLM_PROVIDER must be 'fake' or 'openai', "
+                f"got {cls.AGENT_LLM_PROVIDER!r}"
+            )
+        if cls.AGENT_LLM_PROVIDER == "openai" and (
+            not cls.LLM_API_KEY or not cls.LLM_BASE_URL
+        ):
+            raise ValueError(
+                "AGENT_LLM_PROVIDER=openai requires LLM_API_KEY and LLM_BASE_URL"
+            )
         if cls.AGENT_BACKEND != "fastgpt":
             return
         required = [

src/core/fastgpt_client.py

@@ -11,6 +11,7 @@ from .config import Config
 
 _fastgpt_client: AsyncChatClient | None = None
 _chat_backend: ChatBackend | None = None
+_langgraph_graph = None
 
 
 @asynccontextmanager
@@ -22,7 +23,11 @@ async def lifespan(app: FastAPI):
     if Config.AGENT_BACKEND == "langgraph":
         # 作业 2：LangGraph fake 模式不要求 FastGPT 密钥。
         # TODO(作业 2): 若语音仍启用，需在此单独初始化并校验 FastGPT 配置。
-        _chat_backend = LangGraphBackend(build_graph())
+        from ..agent.llm import build_llm_service
+
+        global _langgraph_graph
+        _langgraph_graph = build_graph(llm=build_llm_service())
+        _chat_backend = LangGraphBackend(_langgraph_graph)
     else:
         _fastgpt_client = AsyncChatClient(
             api_key=Config.FASTGPT_API_KEY,
@@ -39,11 +44,23 @@ async def lifespan(app: FastAPI):
         yield
     finally:
         _chat_backend = None
+        _langgraph_graph = None
         if _fastgpt_client:
             await _fastgpt_client.__aexit__(None, None, None)
         _fastgpt_client = None
 
 
+async def delete_chat_session(thread_id: str) -> None:
+    """langgraph 模式下删除会话：清理图 checkpoint（对照计划作业 6 的
+    完整语义——业务字段、事件、路由持久化清理在接入 repository 后补齐）。
+    """
+    if _langgraph_graph is None:
+        raise RuntimeError("LangGraph backend not initialized")
+    checkpointer = getattr(_langgraph_graph, "checkpointer", None)
+    if checkpointer is not None and hasattr(checkpointer, "delete_thread"):
+        checkpointer.delete_thread(thread_id)
+
+
 def get_fastgpt_client() -> AsyncChatClient:
     """Get the initialized FastGPT client."""
     if _fastgpt_client is None: