fix(agent): send stable session headers to OpenCode

app/agent/llm/helper.py

@@ -7,6 +7,7 @@ import time
 from functools import wraps
 from typing import TYPE_CHECKING, Any, List, Optional
 from urllib.parse import urlsplit
+from uuid import uuid4
 
 from langchain_core.messages import AIMessage, AIMessageChunk
 
@@ -881,11 +882,17 @@ class LLMHelper:
             default_headers: dict[str, str] | None,
             model_kwargs: dict[str, Any],
     ) -> tuple[dict[str, str] | None, dict[str, Any]]:
-        """为 OpenAI 与 xAI 官方端点构造稳定提示词缓存路由参数。"""
+        """为官方端点构造稳定会话及提示词缓存路由参数。"""
         cache_key = str(prompt_cache_key or "").strip()
         headers = dict(default_headers or {})
         kwargs = dict(model_kwargs)
         provider_name = str(provider or "").strip().lower()
+        if cls._matches_endpoint_host(base_url, "opencode.ai"):
+            # 主对话沿用脱敏缓存键；独立测试、摘要调用的标识在模型实例内复用。
+            headers["x-opencode-session"] = cache_key or f"moviepilot-{uuid4().hex}"
+            if not any(key.lower() == "user-agent" for key in headers):
+                headers["User-Agent"] = "MoviePilot"
+            return headers, kwargs
         if not cache_key:
             return headers or None, kwargs
 

tests/test_llm_opencode_session.py

@@ -0,0 +1,101 @@
+"""OpenCode 官方端点的会话路由请求头回归测试。"""
+
+import asyncio
+from unittest.mock import AsyncMock
+
+import httpx
+import pytest
+
+from app.agent.llm import helper
+from app.agent.llm.helper import LLMHelper
+
+
+@pytest.mark.parametrize("base_url", [
+    "https://opencode.ai/zen/v1",
+    "https://opencode.ai/zen/go/v1",
+])
+@pytest.mark.parametrize("cache_key", ["moviepilot-agent-private-hash", None])
+def test_opencode_model_sends_stable_session_on_every_request(monkeypatch, base_url, cache_key):
+    """主对话与独立调用经过真实 SDK 后仍携带稳定标识，工具绑定不丢失请求头。"""
+    requests = []
+
+    def respond(request):
+        """在本地捕获 SDK 请求并返回最小聊天响应，禁止真实出站。"""
+        requests.append(request)
+        return httpx.Response(200, json={
+            "id": "chatcmpl-test",
+            "object": "chat.completion",
+            "created": 0,
+            "model": "test-model",
+            "choices": [{"index": 0, "message": {"role": "assistant", "content": "OK"},
+                         "finish_reason": "stop"}],
+        })
+
+    transport = httpx.MockTransport(respond)
+    with httpx.Client(transport=transport) as client:
+        async def invoke():
+            """使用离线运行时创建模型，并验证同步与异步请求使用同一标识。"""
+            async with httpx.AsyncClient(transport=transport) as async_client:
+                monkeypatch.setattr(
+                    helper, "_build_httpx_client",
+                    lambda _proxy, **kwargs: async_client if kwargs.get("async_client") else client,
+                )
+                runtime = AsyncMock()
+                runtime.resolve_runtime.return_value = {
+                    "runtime": "openai_compatible", "model_id": "test-model",
+                    "api_key": "test-key", "base_url": base_url,
+                }
+                model = await LLMHelper.get_llm(
+                    provider="opencode", model="test-model", user_agent="",
+                    use_proxy=False, api_protocol="chat_completions", web_search_mode="disabled",
+                    prompt_cache_key=cache_key, provider_runtime=runtime,
+                )
+                bound = model.bind_tools([{
+                    "type": "function", "function": {"name": "example", "description": "测试工具",
+                        "parameters": {"type": "object", "properties": {}}},
+                }])
+                assert bound.invoke("hello").content == "OK"
+                assert (await bound.ainvoke("again")).content == "OK"
+
+        asyncio.run(invoke())
+
+    assert len(requests) == 2
+    session = requests[0].headers["x-opencode-session"]
+    assert session
+    assert requests[1].headers["x-opencode-session"] == session
+    if cache_key:
+        assert session == cache_key
+    assert requests[0].headers["user-agent"] == "MoviePilot"
+
+
+@pytest.mark.parametrize("base_url", [
+    "https://opencode.ai/zen/go/v1", "https://OPENCODE.AI/zen/v1",
+])
+def test_opencode_independent_models_have_distinct_sessions(base_url):
+    """没有对话标识的独立模型互不共用会话，并保留自定义 UA 和模型参数。"""
+    original = {"user-agent": "custom-client/1.0", "X-Test": "value"}
+    options = {"extra_body": {"test": True}}
+    headers = [LLMHelper._build_openai_prompt_cache_options(
+        provider="custom", base_url=base_url, use_responses_api=True,
+        prompt_cache_key=None, default_headers=original, model_kwargs=options,
+    ) for _ in range(2)]
+    assert headers[0][0]["x-opencode-session"] != headers[1][0]["x-opencode-session"]
+    assert headers[0][0]["user-agent"] == "custom-client/1.0"
+    assert "User-Agent" not in headers[0][0]
+    assert headers[0][0]["X-Test"] == "value"
+    assert headers[0][1] == options
+    assert "x-opencode-session" not in original
+
+
+@pytest.mark.parametrize("base_url", [
+    "https://opencode.ai.example/zen/go/v1", "https://proxy.example/opencode.ai",
+    "https://opencode.ai@proxy.example/v1", "https://[invalid", None,
+])
+def test_opencode_headers_do_not_leak_to_other_hosts(base_url):
+    """供应商名称与路径不能代替官方主机校验，兼容端点保持原有参数。"""
+    headers, kwargs = LLMHelper._build_openai_prompt_cache_options(
+        provider="opencode", base_url=base_url, use_responses_api=False,
+        prompt_cache_key="private-cache-key", default_headers=None, model_kwargs={},
+    )
+    assert headers is None
+    assert kwargs == {}