refactor: unify skills+cli runtime and streamline ag-ui flow

backend/tests/integration/v1/agent/test_sse_flow_live.py

@@ -1,7 +1,6 @@
 from __future__ import annotations
 
 import base64
-import os
 from pathlib import Path
 from uuid import UUID, uuid4
 
@@ -9,27 +8,34 @@ import httpx
 import pytest
 from sqlalchemy import select
 
+from core.config.settings import config
 from core.db.session import AsyncSessionLocal
 from models.agent_chat_message import AgentChatMessage
 from models.agent_chat_session import AgentChatSession
+from schemas.enums import AgentChatMessageRole
 
-BASE_URL = os.getenv("AGENT_LIVE_BASE_URL", "http://localhost:5775")
+BASE_URL = f"http://localhost:{5775}"
 FIXTURE_IMAGE_PATH = (
     Path(__file__).resolve().parents[3] / "fixtures" / "images" / "calendar_text_cn.png"
 )
 
 
+def _require_test_phone() -> str:
+    phone = config.test.phone
+    if not phone:
+        pytest.fail("SOCIAL_TEST__PHONE is required for live integration tests")
+    return phone
+
+
 async def _live_access_token(client: httpx.AsyncClient) -> str:
-    phone = os.getenv("AGENT_LIVE_PHONE")
-    password = os.getenv("AGENT_LIVE_PASSWORD")
-    if not phone or not password:
-        pytest.fail(
-            "AGENT_LIVE_INTEGRATION=1 requires AGENT_LIVE_PHONE and AGENT_LIVE_PASSWORD"
-        )
+    phone = _require_test_phone()
+    if not phone.startswith("+"):
+        phone = f"+{phone}"
+    code = config.test.code or "000000"
 
     response = await client.post(
-        f"{BASE_URL}/api/v1/auth/sessions",
-        json={"phone": phone, "password": password},
+        f"{BASE_URL}/api/v1/auth/phone-session",
+        json={"phone": phone, "token": code},
     )
     response_text = response.text.strip().replace("\n", " ")
     truncated_text = response_text[:200]
@@ -48,8 +54,8 @@ async def _live_access_token(client: httpx.AsyncClient) -> str:
 @pytest.mark.asyncio
 @pytest.mark.live
 async def test_agent_sse_closed_loop_live() -> None:
-    if os.getenv("AGENT_LIVE_INTEGRATION") != "1":
-        pytest.skip("set AGENT_LIVE_INTEGRATION=1 to run live integration test")
+    if config.runtime.environment not in {"dev", "test"}:
+        pytest.skip("live integration tests require dev or test environment")
 
     async with httpx.AsyncClient(timeout=30.0) as client:
         token = await _live_access_token(client)
@@ -67,7 +73,7 @@ async def test_agent_sse_closed_loop_live() -> None:
                 ],
                 "tools": [],
                 "context": [],
-                "forwardedProps": {"agent_type": "worker"},
+                "forwardedProps": {"runtime_mode": "chat"},
             },
         )
         assert run_resp.status_code == 202
@@ -110,8 +116,8 @@ async def test_agent_sse_closed_loop_live() -> None:
 @pytest.mark.asyncio
 @pytest.mark.live
 async def test_agent_runs_events_history_live_with_image_input() -> None:
-    if os.getenv("AGENT_LIVE_INTEGRATION") != "1":
-        pytest.skip("set AGENT_LIVE_INTEGRATION=1 to run live integration test")
+    if config.runtime.environment not in {"dev", "test"}:
+        pytest.skip("live integration tests require dev or test environment")
 
     image_data = base64.b64encode(FIXTURE_IMAGE_PATH.read_bytes()).decode("ascii")
 
@@ -143,7 +149,7 @@ async def test_agent_runs_events_history_live_with_image_input() -> None:
                 ],
                 "tools": [],
                 "context": [],
-                "forwardedProps": {"agent_type": "worker"},
+                "forwardedProps": {"runtime_mode": "chat"},
             },
         )
         assert run_resp.status_code == 202
@@ -221,3 +227,78 @@ async def test_agent_runs_events_history_live_with_image_input() -> None:
         assert user_attachments
         assert isinstance(user_attachments[0], dict)
         assert isinstance(user_attachments[0].get("path"), str)
+
+
+@pytest.mark.asyncio
+@pytest.mark.live
+async def test_agent_tool_call_result_persisted_live() -> None:
+    if config.runtime.environment not in {"dev", "test"}:
+        pytest.skip("live integration tests require dev or test environment")
+
+    thread_id = str(uuid4())
+
+    async with httpx.AsyncClient(timeout=30.0) as client:
+        token = await _live_access_token(client)
+        headers = {"Authorization": f"Bearer {token}"}
+
+        run_resp = await client.post(
+            f"{BASE_URL}/api/v1/agent/runs",
+            headers=headers,
+            json={
+                "threadId": thread_id,
+                "runId": "run-tool-verify-1",
+                "state": {},
+                "messages": [
+                    {
+                        "id": "u1",
+                        "role": "user",
+                        "content": "帮我查一下明天有哪些日程安排",
+                    }
+                ],
+                "tools": [],
+                "context": [],
+                "forwardedProps": {"runtime_mode": "chat"},
+            },
+        )
+        assert run_resp.status_code == 202
+
+        accepted = run_resp.json()
+        assert str(accepted["threadId"]) == thread_id
+
+        events_url = f"{BASE_URL}/api/v1/agent/runs/{thread_id}/events?runId=run-tool-verify-1"
+        event_names: list[str] = []
+        async with client.stream(
+            "GET", events_url, headers=headers, timeout=90.0
+        ) as sse_resp:
+            assert sse_resp.status_code == 200
+            async for line in sse_resp.aiter_lines():
+                if line.startswith("event:"):
+                    event_name = line.split(":", 1)[1].strip()
+                    event_names.append(event_name)
+                    if event_name in {"RUN_FINISHED", "RUN_ERROR"}:
+                        break
+
+        assert "RUN_STARTED" in event_names, (
+            f"missing RUN_STARTED, got: {event_names}"
+        )
+
+        finished_ok = "RUN_FINISHED" in event_names
+        finished_err = "RUN_ERROR" in event_names
+        assert finished_ok or finished_err, (
+            f"no terminal event, got: {event_names}"
+        )
+
+    async with AsyncSessionLocal() as session:
+        rows = await session.execute(
+            select(AgentChatMessage).where(
+                AgentChatMessage.session_id == UUID(thread_id),
+                AgentChatMessage.role == AgentChatMessageRole.TOOL,
+            )
+        )
+        tool_messages = list(rows.scalars().all())
+
+        if finished_ok:
+            assert len(tool_messages) >= 1, (
+                f"expected >=1 role='tool' message but found {len(tool_messages)}. "
+                f"SSE events: {event_names}"
+            )