[Bug] The metadata is null in /process API response #270
Description
Bug Description
The metadata is null in the /process API response using agentscope-runtime v1.0.0
Affected Component
- Engine
- Sandbox
- Tools
- Common
- Documentation
- Other:___________
Reproduction Steps
- Your code:
main.py
import asyncio
from agentscope_runtime.engine.deployers.local_deployer import LocalDeployManager
from agent_app_test import agent_app_test
async def main():
await agent_app_test.deploy(LocalDeployManager(host="127.0.0.1", port=8088))
if __name__ == "__main__":
asyncio.run(main())
input("按 Enter 键停止服务器...")agent_app_test.py
import logging
import os
from agentscope.agent import ReActAgent
from agentscope.model import DashScopeChatModel
from agentscope.formatter import DashScopeChatFormatter
from agentscope.tool import Toolkit, execute_python_code
from agentscope.pipeline import stream_printing_messages
from agentscope_runtime.engine import AgentApp
from agentscope_runtime.engine.schemas.agent_schemas import AgentRequest
from agentscope_runtime.adapters.agentscope.memory import (
AgentScopeSessionHistoryMemory,
)
from agentscope_runtime.engine.services.agent_state import (
InMemoryStateService,
)
from agentscope_runtime.engine.services.session_history import (
InMemorySessionHistoryService,
)
from agentscope.rag import SimpleKnowledge
from agentscope.embedding import DashScopeTextEmbedding
from milvus import MilvusStore
from dotenv import load_dotenv
from pydantic import BaseModel, Field
print("✅ 依赖导入成功")
load_dotenv()
print(f"api_key:{os.getenv('DASHSCOPE_API_KEY')}")
agent_app_test = AgentApp(
app_name="Friday",
app_description="A helpful assistant",
)
print("✅ AgentApp created successfully")
class RuleModel(BaseModel):
left: str = Field(description="Left Field")
operator: str = Field(
description="operator, example:>,<,>=,<=,==,contains",
)
right: str = Field(description="Right Field")
score: float = Field(description="rule score")
@agent_app_test.init
async def init_func(self):
self.state_service = InMemoryStateService()
self.session_service = InMemorySessionHistoryService()
await self.state_service.start()
await self.session_service.start()
@agent_app_test.shutdown
async def shutdown_func(self):
await self.state_service.stop()
await self.session_service.stop()
@agent_app_test.query(framework="agentscope")
async def query_func(
self,
msgs,
request: AgentRequest = None,
**kwargs,
):
session_id = request.session_id
user_id = request.user_id
state = await self.state_service.export_state(
session_id=session_id,
user_id=user_id,
)
toolkit = Toolkit()
agent = ReActAgent(
name="Friday",
model=DashScopeChatModel(
"qwen-max",
api_key=os.getenv("DASHSCOPE_API_KEY"),
),
sys_prompt="# Identity And Core Mission"
"You are a specialized rule generation assistant designed to create decision rules in a specific JSON format."
"Your core function is to analyze user descriptions of business rules and convert them into standardized rule structures."
"## Instructions"
"1. **Determine the Left Field:** Identify the left-hand parameter of the rule based on the user's input."
"2. **Determine the Operator:** Identify the operator based on the user's input. The operator must be one of the following list: >, <, >=, <=, ==, contains. Assign it to operator."
"3. **Determine the Right Field:** Identify the right-hand parameter of the rule based on the user's input and judge whether it is a constant:"
"* If it is a constant, directly assign it to right."
"* If it is not a constant, search the knowledge base for the most similar field name and assign it to right."
"4. **Determine the Score:** Determine the score based on the user's input:"
"* If the user explicitly specifies a score, set it to the specified value."
"* If the user describes a rejection intent, set the score to 60."
"## Output Specification"
"### Format Requirements"
"You should only output a JSON object in the following:"
"```json"
"{"
"\"left\": \"身份证关联手机号个数\","
"\"operator\": \">\","
"\"right\": \"2\","
"\"score\": 60"
"}"
"```"
"### Field Definitions"
"- **`left`**: Target field identifier (mapped from user description)"
"- **`operator`**: One of: `>`, `<`, `>=`, `<=`, `==`, `contains`"
"- **`right`**: Reference value (constant) or comparative field name"
"- **`score`**: Action indicator → `60` for rejection, `40` for manual review",
toolkit=toolkit,
memory=AgentScopeSessionHistoryMemory(
service=self.session_service,
session_id=session_id,
user_id=user_id,
),
formatter=DashScopeChatFormatter(),
)
agent.set_console_output_enabled(enabled=True)
if state:
agent.load_state_dict(state)
async for msg, last in stream_printing_messages(
agents=[agent],
coroutine_task=agent(msgs,structured_model=RuleModel),
# coroutine_task=agent(msgs),
):
yield msg, last
state = agent.state_dict()
await self.state_service.save_state(
user_id=user_id,
session_id=session_id,
state=state,
)
print("✅ 智能体和端点配置成功")- How to run: python main.py
- Error occurs at:
curl --location --request POST 'http://localhost:8088/process'
--header 'Content-Type: application/json'
--data-raw '{
"input": [
{
"role": "user",
"content": [
{"type": "text", "text": "如果7天内关联的设备ID大于1个，那么就拒绝"}
]
}
],
"session_id": "test_session_001",
"user_id": "test_user_001"
}'
Expected vs Actual Behavior
Expected: the metadata in the response is not null
Actual: the metadata in the response is null
Error Messages
Environment
- **AgentScope Runtime Version:v1.0.0
- **Python Version:3.11
- **OS:macOS 13.0.1
- **Installation: pip