腾讯云LKE智能体对话

腾讯云智能体开发平台 HTTP SSE 对话接口技能。用于调用腾讯云智能体对话接口，支持流式响应处理、事件解析（reply/token_stat/reference/error/thought）。在需要与腾讯云LKE智能体进行HTTP SSE对话时使用，包括发送消息、处理流式响应、解析各类事件。

Security Scan

The skill is an instruction-and-example client for Tencent Cloud LKE HTTP SSE chat and the included docs and script match the described purpose with no disproportionate or hidden requirements.

Purpose & Capability

Name/description, SKILL.md, reference docs, and the provided Python script all describe the same HTTP SSE chat client for Tencent LKE and require only an AppKey and request parameters; there are no unrelated credentials, binaries, or capabilities requested.

Instruction Scope

SKILL.md and the example code limit themselves to constructing POST requests to the Tencent SSE endpoint and parsing SSE responses. The instructions do not ask the agent to read unrelated files, system credentials, or external endpoints beyond the documented Tencent host. One note: the AppKey is passed as a CLI argument in the sample script, which is sensitive and may be exposed via process listings—handle secrets carefully.

Install Mechanism

There is no install spec and the skill is instruction-only with an example script. The included requirements note recommends installing requests or the Tencent SDK via pip (standard, expected). No remote downloads or archive extraction are present.

Credentials

The skill declares no required environment variables or config paths. The only secret needed in practice is the Tencent AppKey (passed as a CLI argument in the example). This is proportional to the stated purpose, but the user should avoid exposing the AppKey in process arguments or logs.

Persistence & Privilege

The skill does not request always:true and does not modify other skills or system-wide settings. It runs on demand and requires no persistent elevated privileges.

Assessment

This skill appears coherent and implements a Tencent LKE SSE chat client. Before installing or running: verify you trust the skill source (source/homepage are unknown), keep your AppKey secret (avoid passing it on the command line where other local users can see it), and consider reviewing the included script yourself. If you plan to run this on shared infrastructure, modify the script to read the AppKey from a secure environment variable or prompt rather than a visible CLI argument. If you need stronger assurance, request the publisher's origin or compare the endpoint/behavior against official Tencent Cloud docs.

腾讯云智能体对话接口 (HTTP SSE)

本技能提供腾讯云智能体开发平台的HTTP SSE对话接口调用指导。

接口概览

• 请求地址: https://wss.lke.cloud.tencent.com/v1/qbot/chat/sse
• 请求方式: POST
• 协议: HTTP SSE (Server-Sent Events)

快速开始

1. 获取 AppKey

在腾讯云控制台 > 智能体开发平台 > 应用管理 > 发布管理中获取应用的 AppKey。

2. 基本调用示例

curl --location 'https://wss.lke.cloud.tencent.com/v1/qbot/chat/sse' \
--header 'Content-Type: application/json' \
--data '{
  "session_id": "a29bae68-cb1c-489d-8097-6be78f136acf",
  "bot_app_key": "YourAppKey",
  "visitor_biz_id": "a29bae68-cb1c-489d-8097-6be78f136acf",
  "content": "你好",
  "incremental": true,
  "streaming_throttle": 10
}'

请求参数

参数名	类型	必填	说明
session_id	string(64)	是	会话ID，2-64字符，格式：^[a-zA-Z0-9_-]{2,64}$
bot_app_key	string(128)	是	应用密钥
visitor_biz_id	string(64)	是	访客ID，标识当前用户
content	string	是	消息内容
request_id	string(255)	否	请求ID，用于消息串联
file_infos	Object[]	否	文件信息（需配合实时文档解析）
streaming_throttle	int32	否	流式回包频率控制，默认5
custom_variables	map[string]string	否	自定义参数（工作流/API参数）
system_role	string	否	角色指令（提示词）
incremental	bool	否	内容是否增量输出，默认false
search_network	string	否	联网搜索：enable/disable/空（跟随配置）
model_name	string	否	指定模型：hunyuan/lke-deepseek-r1/lke-deepseek-v3 等
stream	string	否	是否流式：enable/disable/空
workflow_status	string	否	工作流开关：enable/disable/空

file_infos 数据结构

{
  "file_name": "文档.pdf",
  "file_size": "1024000",
  "file_url": "https://cos.url/xxx",
  "file_type": "pdf",
  "doc_id": "doc_xxx"
}

响应事件类型

SSE流会返回以下事件类型：

1. reply - 回复事件

{
  "type": "reply",
  "payload": {
    "content": "回复内容",
    "is_final": false,
    "is_evil": false,
    "is_from_self": false,
    "is_llm_generated": true,
    "record_id": "msg_xxx",
    "session_id": "session_xxx",
    "timestamp": 1701330805,
    "reply_method": 1,
    "knowledge": [...],
    "quote_infos": [...]
  }
}

关键字段:

• is_final: true表示消息输出完毕
• is_evil: true表示命中敏感内容
• reply_method: 回复方式（1=大模型回复, 16=工作流回复, 18=智能体回复等）
• quote_infos: 引用信息，用于标注参考来源

2. token_stat - Token统计事件

{
  "type": "token_stat",
  "payload": {
    "elapsed": 1616,
    "token_count": 323,
    "status_summary": "success",
    "procedures": [
      {
        "name": "knowledge",
        "title": "调用知识库",
        "status": "success",
        "input_count": 308,
        "output_count": 15
      }
    ]
  }
}

3. reference - 参考来源事件

{
  "type": "reference",
  "payload": {
    "record_id": "msg_xxx",
    "references": [
      {
        "id": "ref_xxx",
        "type": 2,
        "name": "文档名称",
        "doc_name": "xxx.docx",
        "url": "https://..."
      }
    ]
  }
}

参考来源类型:

• 1: 问答
• 2: 文档片段
• 4: 联网检索内容

4. thought - 思考事件 (DeepSeek-R1)

{
  "type": "thought",
  "payload": {
    "procedures": [
      {
        "name": "thought",
        "title": "思考",
        "debugging": {
          "content": "思考过程..."
        }
      }
    ]
  }
}

5. error - 错误事件

{
  "type": "error",
  "error": {
    "code": 460004,
    "message": "应用不存在"
  }
}

错误码

错误码	说明
400	请求参数错误
460001	Token校验失败
460004	应用不存在
460007	会话创建失败
460011	超出并发数限制
460020	模型请求超时
460032	模型余额不足
460034	输入内容过长

进阶用法

使用自定义模型

{
  "model_name": "lke-deepseek-r1",
  "content": "你好"
}

支持的模型:

• hunyuan: 混元高级版
• hunyuan-turbo: 混元Turbo版
• lke-deepseek-r1: DeepSeek-R1
• lke-deepseek-v3: DeepSeek-V3
• lke-deepseek-r1-0528: DeepSeek-R1-0528
• lke-deepseek-v3-0324: DeepSeek-V3-0324

工作流参数传递

{
  "custom_variables": {
    "UserID": "10220022",
    "Data": "{\"Score\":{\"Chinese\":89}}"
  }
}

知识库检索范围设置

{
  "custom_variables": {
    "tag_field": "user1|user2"
  }
}

前端渲染组件

如需前端渲染消息，可使用官方组件:

# Vue 2 或 React
npm install lke-component

# Vue 3
npm install lke-component-vue3

参考资料

• 完整接口文档: references/api_reference.md
• 示例脚本: scripts/