Skip to content
能力中心
产品中心
应用市场
WebOffice
开发者后台

创建对话

一次对话包含用户问和智能体回答两个动作

标签对话

请求说明

字段
请求地址
https://openapi.wps.cn/v7/devhub/app/{app_id}/chat/create
HTTP 方法
POST
接口描述
开始对话
签名方式
KSO-1
限频策略
权限要求
智能体对话管理(用户授权) kso.devhub_chat.readwrite

路径参数 (Path)

属性名类型是否必填描述可选值
app_id
string-

请求体(Body)

请求体格式: application/json

没有可用的数据

请求体示例

json
{
  "files": [
    "string"
  ],
  "session_id": "string",
  "stream": true,
  "text": "string",
  "variables": [
    {
      "key": "string",
      "value": "string"
    }
  ]
}

响应体(Response)

HTTP状态码: 200
响应体格式: application/json

没有可用的数据

响应体示例

json
{
  "data": {
    "chat_id": "string",
    "session_id": "string"
  },
  "code": 0,
  "msg": "string"
}

非流式对话响应体

名称参数类型说明
codeinteger响应代码。非 0 表示失败,参照《状态码说明》
msgstring响应信息
dataobject响应数据
∟ session_idstring会话id,请求参数中如果没传会话id则会新生成一个会话,并返回id
∟ chat_idstring对话id

非流式响应体示例

json
{
    "code": 0,
    "msg": "success",
    "data": {
        "session_id":"xxxxxxxxx",
        "chat_id":"xxxxxxxxxxx"
    }
}

流式响应

响应结果是多个流式消息,每个流式消息结构体如下

{"event": "end", "data": {}}

event代表事件类型,枚举如下

  • vertices_sorted运行前对所有节点排序
  • add_message 流程中一个节点运行输出的信息
  • end_vertex 流程中一个节点运行结束
  • token 智能体流式返回
  • end 整个流程结束

data代表事件数据,具体数据样例如下

{"event": "add_message", "data": {"timestamp": "2025-05-09T15:03:47.532881", "sender": "Machine", "sender_name": "Agent", "session_id": "7038a190-c4e9-44c3-9305-f560266be942", "text": "\u4e2d\u56fd\u7684\u603b\u9762\u79ef\u5927\u7ea6\u4e3a960\u4e07\u5e73\u65b9\u516c\u91cc\uff0c\u662f\u4e16\u754c\u4e0a\u9762\u79ef\u7b2c\u4e09\u6216\u7b2c\u56db\u5927\u7684\u56fd\u5bb6\uff0c\u5177\u4f53\u6392\u540d\u53d6\u51b3\u4e8e\u5bf9\u6709\u4e89\u8bae\u5730\u533a\u7684\u7edf\u8ba1\u65b9\u5f0f\u3002\u8fd9\u4f7f\u5f97\u4e2d\u56fd\u5728\u4e9a\u6d32\u662f\u9762\u79ef\u6700\u5927\u7684\u56fd\u5bb6\uff0c\u5e76\u4e14\u6709\u7740\u975e\u5e38\u4e30\u5bcc\u548c\u591a\u6837\u7684\u5730\u7406\u666f\u89c2\u3002", "files": [], "error": false, "edit": false, "properties": {"text_color": null, "background_color": null, "edited": false, "source": {"id": null, "display_name": null, "source": null}, "icon": "Bot", "allow_markdown": false, "positive_feedback": null, "state": "complete", "targets": []}, "category": "message", "content_blocks": [{"title": "Agent Steps", "contents": [{"type": "text", "duration": 18, "header": {"title": "Input", "icon": "MessageSquare"}, "text": "**Input**: \u4f60\u77e5\u9053\u4e2d\u56fd\u6709\u591a\u5927\u4e48"}, {"type": "text", "duration": 33, "header": {"title": "Input", "icon": "MessageSquare"}, "text": "**Input**: \u4f60\u77e5\u9053\u4e2d\u56fd\u6709\u591a\u5927\u4e48"}, {"type": "text", "duration": 0, "header": {"title": "Input", "icon": "MessageSquare"}, "text": "**Input**: \u4f60\u77e5\u9053\u4e2d\u56fd\u6709\u591a\u5927\u4e48"}, {"type": "text", "duration": 4012, "header": {"title": "Output", "icon": "MessageSquare"}, "text": "\u4e2d\u56fd\u7684\u603b\u9762\u79ef\u5927\u7ea6\u4e3a960\u4e07\u5e73\u65b9\u516c\u91cc\uff0c\u662f\u4e16\u754c\u4e0a\u9762\u79ef\u7b2c\u4e09\u6216\u7b2c\u56db\u5927\u7684\u56fd\u5bb6\uff0c\u5177\u4f53\u6392\u540d\u53d6\u51b3\u4e8e\u5bf9\u6709\u4e89\u8bae\u5730\u533a\u7684\u7edf\u8ba1\u65b9\u5f0f\u3002\u8fd9\u4f7f\u5f97\u4e2d\u56fd\u5728\u4e9a\u6d32\u662f\u9762\u79ef\u6700\u5927\u7684\u56fd\u5bb6\uff0c\u5e76\u4e14\u6709\u7740\u975e\u5e38\u4e30\u5bcc\u548c\u591a\u6837\u7684\u5730\u7406\u666f\u89c2\u3002"}], "allow_markdown": true, "media_url": null}], "id": "a550de06-0198-4a57-a5cd-75000522bf1d", "flow_id": "72cdefc1-940c-428d-bd8e-198386f3de5c"}}

{"event": "end", "data": {}}

流式对话中获取智能体回复方法如下:

  • 监听event是add_message类型,且data中"sender": "Machine","sender_name": "Agent"的事件,data中的id是消息id
  • 获取事件中的data.text内容即为智能体流式回复
  • 消息中存在chat_id用于后续取消对话