基于 AgentScope x AI Agent A2Z部署平台的生产级别Agent上线Live实战分享

`
    response_msg = await agent.reply(msg)
    print (f"Agent Reply: response {response_msg}")
    response_content = response_msg.content
    print (f"Agent Reply: response response_content {response_content}")

@app.post("/chat")
async def chat(request: ChatRequest) -> StreamingResponse:
    """ Convert the DeepResearch Agent to Live Service."""
    global tavily_search_client, agent
    if agent is None:
        raise HTTPException(status_code=503, detail="Agent not initialized")

    try:
        messages = request.messages
        user_query = ""
        try:
            user_query = messages[-1]["content"] if len(messages) > 0 else "hello"
        except Exception as e1:
            print (f"Failed to process input messages: {e1}")
        print (f"DEBUG: user: {user_query}")
        user_name = "USER_"
        ## set to 1 iteration for example
        msg = Msg(
            user_name,
            content=user_query,
            role="user"
        )

        return StreamingResponse(
            stream_generator(agent, msg),
            media_type="application/json",
        )

    except Exception as e:
        raise HTTPException(status_code=500, detail=f"Failed to chat: {str(e)}") from e
  
async def stream_generator(agent, msg):
    """
    Generator for Chat Messages Assemble AgentScope Response

    message_type: "user", "assistant"
    output_format: "text", "html"
    content_type: "text/markdown", mime-type

    SECTION_THINK = "think"
    SECTION_ANSWER = "answer"
    SECTION_TOOL = "tool"
    SECTION_SYSTEM_MSG = "system_msg"
    SECTION_CONTEXT = "context"

    TEMPLATE_REASONING_HTML = "reason_html"
    TEMPLATE_STREAMING_CONTENT_TYPE = "streaming_content_type"
    """
    message_type = "assistant"
    output_format = "text"
    content_type = "text/markdown"
    section = "answer"
    streaming_separator = "\n"
    TEMPLATE_STREAMING_CONTENT_TYPE = "streaming_content_type"

    ## Initial Chunk
    initial_chunk = json.dumps(assembly_message(message_type, output_format, "DeepResearch Task Starting...", content_type=content_type, section=section, message_id= str(uuid.uuid4()), template=TEMPLATE_STREAMING_CONTENT_TYPE) )
    yield initial_chunk + streaming_separator

    ## result is a message class Msg
    response_msg = await agent.reply(msg)
    print (f"Agent Reply: response {response_msg}")
    response_content = response_msg.content
    print (f"Agent Reply: response response_content {response_content}")

    output_message_id = response_msg.id
    content_type_chunk = json.dumps(assembly_message(message_type, output_format, response_content, content_type=content_type, section=section, message_id=output_message_id, template=TEMPLATE_STREAMING_CONTENT_TYPE) )

    print (f"stream_generator response Result: {response_msg}")
    yield content_type_chunk + streaming_separator

export DASHSCOPE_API_KEY="your_dashscope_api_key_here"
export TAVILY_API_KEY="your_tavily_api_key_here"

INFO:     Started server process [1]
INFO:     Waiting for application startup.
Application startup...
Tavily MCP server running on stdio
2026-02-11 04:00:19,324 | INFO    | _stateful_client_base:connect:66 - MCP client connected.
Lifecycle closed at the end...
INFO:     Application startup complete.
> SUCCESS: ✅ Deployment Complete!

curl -X POST "https://agentscope.aiagenta2z.com/deep_research_agent/chat" \
  -H "Content-Type: application/json" \
  -d '{"messages":[{"role":"user","content":"Calculate 1+1 result"}]}'

{"type": "assistant", "format": "text", "content": "DeepResearch Task Starting...", "section": "answer", "message_id": "202d21fd-c71b-4a11-ba35-e2cb3c7d5947", "content_type": "text/markdown", "template": "streaming_content_type", "task_ids": "", "tab_message_ids": "", "tab_id": ""}
{"type": "assistant", "format": "text", "content": "{\n  \"type\": \"text\",\n  \"text\": \"Overwrite /app/agentscope_examples/deep_research_agent/deepresearch_agent_demo_env/Friday260211040019_detailed_report.md successfully.\\n# Calculation of 1 + 1: A Foundational Arithmetic Operation\\n\\n## Step 1: Confirming the Context of the Operation\\n\\nThe expression \\\"1 + 1\\\" is interpreted within the standard framework of elementary arithmetic unless otherwise specified. In this context:\\n\\n- The numerals \\\"1\\\" represent the natural number one, which is the first positive integer in the set \u2115 = {1, 2, 3, ...} (or sometimes defined to include 0, depending on convention).\\n- The symbol \\\"+\\\" denotes the binary operation of addition as defined in the Peano axioms for natural numbers or as commonly taught in primary education.\\n- The numeral system assumed is base-10 (decimal), which is the standard positional numeral system used globally for everyday arithmetic.\\n\\nNo alternative interpretations\u2014such as those from Boolean logic, modular arithmetic, or abstract algebra\u2014are indicated in the subtask, so we proceed under the assumption of classical arithmetic in the natural numbers.\\n\\n## Step 2: Performing the Calculation\\n\\nUsing the definition of addition for natural numbers:\\n\\n- By the successor function in Peano arithmetic, the number 2 is defined as the successor of 1, denoted S(1).\\n- Addition is recursively defined such that:\\n  - \\\\( a + 0 = a \\\\)\\n  - \\\\( a + S(b) = S(a + b) \\\\)\\n\\nThus:\\n\\\\[\\n1 + 1 = 1 + S(0) = S(1 + 0) = S(1) = 2\\n\\\\]\\n\\nAlternatively, from empirical and educational foundations:\\n- Counting one object and then adding another yields a total of two objects.\\n- This is consistent across physical, symbolic, and computational representations.\\n\\nTherefore, **1 + 1 = 2**.\\n\\n## Step 3: Validation\\n\\nThis result is universally accepted in standard mathematics and has been formally verified in foundational logical systems. Notably:\\n\\n- In *Principia Mathematica* by Alfred North Whitehead and Bertrand Russell (1910\u20131913), the proposition \\\"1 + 1 = 2\\\" is rigorously derived from set-theoretic and logical axioms. It appears as Proposition \u221754.43 in Volume I, with the actual proof completed in Volume II after hundreds of pages of preliminary logic. While famously taking over 300 pages to reach, this underscores the depth of formal verification possible\u2014even for seemingly trivial statements.\\n\\n- Modern computational systems (e.g., calculators, programming languages like Python, MATLAB, or Wolfram Language) all return `2` when evaluating `1 + 1`.\\n\\n- Educational curricula worldwide introduce this as the first non-trivial addition fact, reinforcing its role as a cornerstone of numerical literacy.\\n\\n## Conclusion\\n\\nUnder standard arithmetic in the base-10 numeral system, using the conventional meaning of numerals and the addition operator, the expression **1 + 1 evaluates unequivocally to 2**. This result is mathematically sound, logically consistent, empirically verifiable, and computationally confirmed.\\n\\n**Final Answer: 2**\"\n}", "section": "answer", "message_id": "My5UpF5iRxcWbyooMHqogZ", "content_type": "text/markdown", "template": "streaming_content_type", "task_ids": "", "tab_message_ids": "", "tab_id": ""}