feat(chat): 推荐问法 API + Agent 软超时默认 45s

- GET /api/py/chat/suggested-questions 供 Ink BFF 转发
- AGENT_MAX_LATENCY_MS 未设时默认 45000，降低 Intent 慢于 15s 即 V1 降级的误触发

Co-authored-by: Cursor <cursoragent@cursor.com>

Author: CyningMM

api/agent.py

@@ -85,7 +85,7 @@ def __init__(
         self._tools = {t.name: t for t in tools}
         self._memory = memory
         self._max_steps = max(1, int(os.getenv("AGENT_MAX_STEPS", "5")))
-        self._max_latency_ms = max(1000, int(os.getenv("AGENT_MAX_LATENCY_MS", "15000")))
+        self._max_latency_ms = max(1000, int(os.getenv("AGENT_MAX_LATENCY_MS", "45000")))
         self._min_confidence = float(os.getenv("INTENT_MIN_CONFIDENCE", "0.6"))
 
     def _select_tool(self, tool_name: ToolName) -> Tool:

api/index.py

@@ -1142,3 +1142,20 @@ async def py_admin_ingest(
             status = 400
         return JSONResponse({"ok": False, "error": msg}, status_code=status)
 
+
+@app.get("/api/py/chat/suggested-questions")
+def chat_suggested_questions() -> JSONResponse:
+    """返回推荐问法列表，供前端展示（不再本地写死）。"""
+    return JSONResponse(
+        {
+            "ok": True,
+            "questions": [
+                "《AI 编程可闭环协作》卷三讲什么？Harness 和签收是什么？",
+                "Tech Graph 是什么",
+                "冷/温/热 和 架构三层 区别？",
+                "简单介绍下刘新宁",
+                "AI Ink Brain 的架构是怎样的",
+            ],
+        }
+    )
+

tests/test_chat_suggested_questions_route.py

@@ -0,0 +1,25 @@
+from __future__ import annotations
+
+import importlib
+
+import pytest
+from fastapi.testclient import TestClient
+
+
+def test_chat_suggested_questions_returns_list(monkeypatch: pytest.MonkeyPatch) -> None:
+    monkeypatch.setenv("SYNC_ADMIN_SECRET", "secret-token-1234567890")
+    monkeypatch.setenv("CHATBI_USE_AGENT", "false")
+
+    index = importlib.import_module("api.index")
+    importlib.reload(index)
+
+    client = TestClient(index.app)
+    res = client.get("/api/py/chat/suggested-questions")
+
+    assert res.status_code == 200
+    data = res.json()
+    assert data["ok"] is True
+    questions = data["questions"]
+    assert isinstance(questions, list)
+    assert len(questions) >= 1
+    assert all(isinstance(q, str) and q.strip() for q in questions)