fix(chat): simplify header cards and llm fallback

agent_core/llm_provider.py

@@ -1,7 +1,7 @@
 from dataclasses import dataclass
 import json
 import os
-from urllib.error import URLError
+from urllib.error import HTTPError, URLError
 from urllib.request import Request, urlopen
 
 
@@ -75,12 +75,28 @@ class OpenAICompatibleProvider:
         if response_format:
             payload["response_format"] = response_format
         try:
-            data = _post_json(
+            try:
-                base_url=self.base_url,
+                data = _post_json(
-                endpoint="chat/completions",
+                    base_url=self.base_url,
-                api_key=self.api_key,
+                    endpoint="chat/completions",
-                payload=payload,
+                    api_key=self.api_key,
-            )
+                    payload=payload,
+                )
+            except RuntimeError as exc:
+                # 部分 OpenAI 兼容供应商或模型不支持 response_format。
+                # 保留结构化优先，遇到 400 时退回普通对话，避免演示链路被接口能力差异阻断。
+                if not response_format or "HTTP Error 400" not in str(exc):
+                    raise
+                fallback_payload = {
+                    "model": self.model_name,
+                    "messages": messages,
+                }
+                data = _post_json(
+                    base_url=self.base_url,
+                    endpoint="chat/completions",
+                    api_key=self.api_key,
+                    payload=fallback_payload,
+                )
             choice = data.get("choices", [{}])[0]
             content = choice.get("message", {}).get("content", "")
             return LLMResponse(
@@ -197,5 +213,11 @@ def _post_json(base_url: str, endpoint: str, api_key: str, payload: dict) -> dic
     try:
         with urlopen(request, timeout=60) as response:
             return json.loads(response.read().decode("utf-8"))
+    except HTTPError as exc:
+        error_body = exc.read().decode("utf-8", errors="ignore")
+        error_detail = f"{exc}"
+        if error_body:
+            error_detail = f"{error_detail} {error_body}"
+        raise RuntimeError(f"OpenAI 兼容接口调用失败：{error_detail}") from exc
     except URLError as exc:
         raise RuntimeError(f"OpenAI 兼容接口调用失败：{exc}") from exc

templates/chat/index.html

@@ -359,33 +359,6 @@
     {% endif %}
   </section>
 
-  {% if conversation %}
-    <section class="grid-2">
-      <article class="panel">
-        <h2 class="section-title">顶部对话上下文</h2>
-        <p class="section-copy">进入会话后，先用当前批次、产品和风险状态快速建立审核上下文。</p>
-        <ul class="detail-list">
-          <li class="detail-item"><strong>批次编号</strong><div>{{ conversation_context.batch_id }}</div></li>
-          <li class="detail-item"><strong>产品名称</strong><div>{{ conversation_context.product_name|default:"未识别产品名称" }}</div></li>
-          <li class="detail-item"><strong>当前流程类型</strong><div>{{ conversation_context.workflow_type }}</div></li>
-          <li class="detail-item"><strong>当前审核阶段</strong><div>{{ conversation_context.task_status }}</div></li>
-          <li class="detail-item"><strong>当前最高风险等级</strong><div>{{ conversation_context.highest_risk_level }}</div></li>
-          <li class="detail-item"><strong>是否允许正式导出</strong><div>{{ conversation_context.export_allowed }}</div></li>
-        </ul>
-      </article>
-
-      <article class="panel">
-        <h2 class="section-title">推荐提问模板</h2>
-        <p class="section-copy">用这些提问模板快速进入目录汇总、完整性检查、字段抽取和风险分析。</p>
-        <div class="button-row">
-          {% for item in prompt_templates %}
-            <span class="pill pill-accent">{{ item }}</span>
-          {% endfor %}
-        </div>
-      </article>
-    </section>
-  {% endif %}
-
   <section class="chat-layout">
     <input class="history-toggle" id="history-collapsed" type="checkbox">
     <aside class="panel history-panel">

tests/test_chat.py

@@ -526,7 +526,7 @@ def test_chat_page_shows_upload_entry_and_dynamic_context_cards(client, db):
     assert "飞书通知 / 待处理" in content
 
 
-def test_chat_page_shows_top_context_and_recommended_prompts(client, db):
+def test_chat_page_keeps_prompts_inside_chat_thread_without_top_cards(client, db):
     batch, conversation = _create_conversation_with_batch()
     conversation.task_status = "processing"
     conversation.node_results = [
@@ -545,13 +545,10 @@ def test_chat_page_shows_top_context_and_recommended_prompts(client, db):
 
     content = response.content.decode("utf-8")
     assert response.status_code == 200
-    assert "顶部对话上下文" in content
+    assert "顶部对话上下文" not in content
-    assert "当前流程类型" in content
+    assert "推荐提问模板" not in content
-    assert "registration" in content
+    assert "对话区与节点导航" in content
-    assert "当前审核阶段" in content
     assert "处理中" in content
-    assert "当前最高风险等级" in content
-    assert "推荐提问模板" in content
     assert "请汇总当前资料包的章节点、页数和目录覆盖情况" in content
     assert "请给出当前资料包的高风险项、责任人和整改建议" in content
 
@@ -583,16 +580,12 @@ def test_chat_page_explicitly_shows_batch_product_stage_and_export_context(clien
     assert f"批次：{batch.batch_id}" in content
     assert f"产品：{batch.product_name}" in content
     assert "阶段：处理中" in content
-    assert "批次编号" in content
+    assert "批次编号" not in content
-    assert batch.batch_id in content
+    assert "当前审核阶段" not in content
-    assert "产品名称" in content
+    assert "最高风险等级" in content
-    assert batch.product_name in content
+    assert "高" in content
-    assert "当前审核阶段" in content
-    assert "处理中" in content
-    assert "当前最高风险等级" in content
-    assert ">高<" in content
     assert "是否允许正式导出" in content
-    assert ">否<" in content
+    assert "否" in content
 
 
 def test_chat_page_shows_overview_card_from_conversation_summary(client, db):

tests/test_llm_provider.py

@@ -5,6 +5,8 @@ from agent_core.llm_provider import (
     create_llm_provider,
     get_runtime_llm_config,
 )
+from urllib.error import HTTPError
+from io import BytesIO
 
 
 def test_create_llm_provider_requires_api_key_for_openai_compatible():
@@ -72,6 +74,53 @@ def test_openai_compatible_provider_posts_chat_completion(monkeypatch):
     assert captured["headers"]["Authorization"] == "Bearer sk-test"
 
 
+def test_openai_compatible_provider_falls_back_when_response_format_is_rejected(monkeypatch):
+    captured_bodies = []
+
+    class FakeResponse:
+        def __enter__(self):
+            return self
+
+        def __exit__(self, exc_type, exc, traceback):
+            return False
+
+        def read(self):
+            return b'{"choices":[{"message":{"content":"fallback ok"}}],"model":"demo-model"}'
+
+    def fake_urlopen(request, timeout):
+        body = request.data.decode("utf-8")
+        captured_bodies.append(body)
+        if len(captured_bodies) == 1:
+            raise HTTPError(
+                request.full_url,
+                400,
+                "Bad Request",
+                hdrs=None,
+                fp=BytesIO(b'{"error":{"message":"response_format is not supported"}}'),
+            )
+        return FakeResponse()
+
+    monkeypatch.setattr("agent_core.llm_provider.urlopen", fake_urlopen)
+    provider = create_llm_provider(
+        {
+            "LLM_PROVIDER": "openai_compatible",
+            "LLM_API_KEY": "sk-test",
+            "LLM_BASE_URL": "https://api.siliconflow.cn/v1",
+            "LLM_MODEL": "demo-model",
+        }
+    )
+
+    response = provider.generate(
+        [{"role": "user", "content": "hello"}],
+        response_format={"type": "json_object"},
+    )
+
+    assert response.success is True
+    assert response.content == "fallback ok"
+    assert '"response_format"' in captured_bodies[0]
+    assert '"response_format"' not in captured_bodies[1]
+
+
 def test_embedding_provider_uses_openai_compatible_embeddings(monkeypatch):
     class FakeResponse:
         def __enter__(self):