Automate browser interactions, extract page data, and replay saved browser tasks from a CLI.
Coding
Build QQ bots, channels, Mini Programs, and OAuth 2.0 integrations with route-specific implementation guidance.
What it does
Choose between the official QQ Bot API and the community OneBot ecosystem, then implement the official path from registration through deployment. Guidance covers Bearer token auth, WebSocket events and intents, channel and group messaging, Markdown and Ark cards, Mini Program login, and QQ OAuth 2.0 flows, with Python and JavaScript examples.
When to use it
- Choosing official Bot API or OneBot
- Building a channel or group reply bot
- Adding QQ Mini Program login
- Integrating QQ OAuth 2.0 user authorization
The skill document
QQ 开放平台开发 AI 助手
覆盖 QQ 机器人(官方与社区两条路线)、QQ 频道、QQ 小程序、 开放平台 OAuth2.0 的开发指导。
先选路线:官方 Bot API vs 社区 OneBot 生态
做 QQ 机器人第一个决策是走哪条路线,选错返工代价很大:
| 维度 | 官方 Bot API(本文主体) | 社区 OneBot 生态 |
|---|---|---|
| 合规性 | ✅ 官方支持,不封号 | ⚠️ 协议侧实现,有风控/封号风险 |
| 能力范围 | 频道全功能;群/C2C 需审核开通且受限 | 群/私聊全功能(模拟客户端) |
| 消息限制 | 被动回复为主,主动消息有条数限制 | 无官方限制 |
| 生态现状 | q.qq.com 注册,官方 SDK | OneBot 标准;go-cqhttp 已停止维护,现役实现为 NapCat、Lagrange 等,框架层 NoneBot2/Koishi 仍活跃 |
| 适合 | 对外服务、商业机器人 | 个人自用、内网工具 |
建议:对外产品一律走官方 Bot API;社区路线仅限个人自用且 自担风控风险,本 skill 不提供协议逆向细节。
身份与能力
- 精通 QQ 官方 Bot API,能指导机器人从注册到上线全流程
- 熟练掌握 QQ 频道开发(消息、富文本、Markdown、Embed、Ark 模板)
- 了解 QQ 小程序开发基础与发布流程
- 深入理解 QQ 开放平台 OAuth2.0 授权体系
- 熟悉 WebSocket 事件网关与事件分发机制
QQ 机器人开发
注册与配置
- 访问 QQ 开放平台 注册开发者账号
- 创建机器人应用,获取 AppID 和 AppSecret
- 配置沙箱频道用于测试
- 机器人分为「频道机器人」和「群聊/私聊机器人」两种场景
鉴权方式
QQ 机器人 API 使用 Bearer Token 鉴权,需先获取 access_token:
import requests
def get_qq_bot_token(app_id: str, app_secret: str) -> str:
"""获取 QQ 机器人 access_token"""
url = "https://bots.qq.com/app/getAppAccessToken"
payload = {
"appId": app_id,
"clientSecret": app_secret
}
resp = requests.post(url, json=payload)
data = resp.json()
# access_token 有效期通常为 7200 秒
return data["access_token"]
WebSocket 事件网关
机器人通过 WebSocket 接收实时事件,流程如下:
1. GET /gateway → 获取 WebSocket 连接地址
2. 建立 WebSocket 连接
3. 收到 Hello 事件(opcode=10),获取 heartbeat_interval
4. 发送 Identify 鉴权(opcode=2)
5. 收到 Ready 事件,开始接收消息
6. 定时发送心跳(opcode=1)
import asyncio
import json
import aiohttp
class QQBotWebSocket:
def __init__(self, token: str):
self.token = token
self.api_base = "https://api.sgroup.qq.com"
self.session_id = None
self.seq = None
async def get_gateway(self) -> str:
async with aiohttp.ClientSession() as session:
headers = {"Authorization": f"QQBot {self.token}"}
async with session.get(f"{self.api_base}/gateway", headers=headers) as resp:
data = await resp.json()
return data["url"]
async def connect(self):
gateway_url = await self.get_gateway()
async with aiohttp.ClientSession() as session:
async with session.ws_connect(gateway_url) as ws:
async for msg in ws:
if msg.type == aiohttp.WSMsgType.TEXT:
payload = json.loads(msg.data)
await self.handle_event(ws, payload)
async def handle_event(self, ws, payload: dict):
op = payload.get("op")
if op == 10: # Hello
interval = payload["d"]["heartbeat_interval"]
# 发送 Identify
await ws.send_json({
"op": 2,
"d": {
"token": f"QQBot {self.token}",
"intents": 0 | (1 << 30) | (1 << 1) | (1 << 0),
"shard": [0, 1]
}
})
# 启动心跳
asyncio.create_task(self.heartbeat(ws, interval))
elif op == 0: # Dispatch 事件
self.seq = payload.get("s")
event_type = payload.get("t")
await self.dispatch(event_type, payload.get("d", {}))
async def heartbeat(self, ws, interval_ms: int):
while True:
await asyncio.sleep(interval_ms / 1000)
await ws.send_json({"op": 1, "d": self.seq})
async def dispatch(self, event_type: str, data: dict):
print(f"收到事件: {event_type}, 数据: {data}")
Intents(事件订阅)
Intents 是位掩码,用于声明机器人需要接收的事件类型:
| Intent 值 | 事件类型 | 说明 |
|---|---|---|
1 << 0 | GUILDS | 频道变更(创建/更新/删除) |
1 << 1 | GUILD_MEMBERS | 成员变更(加入/退出) |
1 << 9 | GUILD_MESSAGE_REACTIONS | 表情表态事件 |
1 << 12 | DIRECT_MESSAGE | 频道私信消息 |
1 << 25 | GROUP_AND_C2C_EVENT | 群聊和 C2C 消息事件 |
1 << 30 | AT_MESSAGES | @机器人 消息(公域) |
# 常用 intents 组合
INTENTS_PUBLIC = (1 << 0) | (1 << 1) | (1 << 30) # 频道 + 成员 + @消息
INTENTS_PRIVATE = (1 << 0) | (1 << 1) | (1 << 12) # 含私信
INTENTS_GROUP = (1 << 25) # 群聊和 C2C
消息发送
async def send_channel_message(channel_id: str, content: str, token: str,
msg_id: str = None, image: str = None):
"""向频道子频道发送消息"""
url = f"https://api.sgroup.qq.com/channels/{channel_id}/messages"
headers = {
"Authorization": f"QQBot {token}",
"Content-Type": "application/json"
}
payload = {"content": content}
if msg_id:
payload["msg_id"] = msg_id # 被动回复,引用原消息
if image:
payload["image"] = image # 图片 URL
async with aiohttp.ClientSession() as session:
async with session.post(url, headers=headers, json=payload) as resp:
return await resp.json()
async def send_group_message(group_openid: str, content: str, token: str, msg_id: str):
"""向 QQ 群发送消息(需要 msg_id 被动回复)"""
url = f"https://api.sgroup.qq.com/v2/groups/{group_openid}/messages"
headers = {
"Authorization": f"QQBot {token}",
"Content-Type": "application/json"
}
payload = {
"content": content,
"msg_type": 0, # 0=文本, 2=Markdown, 3=Ark, 4=Embed, 7=富媒体
"msg_id": msg_id
}
async with aiohttp.ClientSession() as session:
async with session.post(url, headers=headers, json=payload) as resp:
return await resp.json()
消息类型
| msg_type | 类型 | 说明 |
|---|---|---|
| 0 | 文本 | 纯文本消息 |
| 2 | Markdown | Markdown 格式消息 |
| 3 | Ark | Ark 模板消息(卡片) |
| 4 | Embed | 嵌入式消息 |
| 7 | 富媒体 | 图片/视频/语音/文件 |
QQ 频道开发
频道管理 API
async def get_guild_info(guild_id: str, token: str) -> dict:
"""获取频道详情"""
url = f"https://api.sgroup.qq.com/guilds/{guild_id}"
headers = {"Authorization": f"QQBot {token}"}
async with aiohttp.ClientSession() as session:
async with session.get(url, headers=headers) as resp:
return await resp.json()
# 返回: {"id", "name", "icon", "owner_id", "member_count", ...}
async def get_channels(guild_id: str, token: str) -> list:
"""获取频道下的子频道列表"""
url = f"https://api.sgroup.qq.com/guilds/{guild_id}/channels"
headers = {"Authorization": f"QQBot {token}"}
async with aiohttp.ClientSession() as session:
async with session.get(url, headers=headers) as resp:
return await resp.json()
子频道类型
| type 值 | 类型 | 说明 |
|---|---|---|
| 0 | 文字子频道 | 发送文本/图片/Ark 消息 |
| 1 | 保留 | - |
| 2 | 语音子频道 | 语音通话 |
| 3 | 保留 | - |
| 4 | 子频道分组 | 分类目录 |
| 10005 | 直播子频道 | 直播推流 |
| 10006 | 应用子频道 | 小程序 |
| 10007 | 论坛子频道 | 帖子讨论 |
Markdown 消息
async def send_markdown_message(channel_id: str, token: str, msg_id: str):
"""发送 Markdown 消息"""
url = f"https://api.sgroup.qq.com/channels/{channel_id}/messages"
headers = {
"Authorization": f"QQBot {token}",
"Content-Type": "application/json"
}
payload = {
"msg_id": msg_id,
"markdown": {
"content": "# 标题\n**加粗文本**\n- 列表项1\n- 列表项2\n[链接](https://example.com)"
}
}
# 也可使用模板 Markdown
# payload = {
# "markdown": {
# "custom_template_id": "模板ID",
# "params": [
# {"key": "title", "values": ["标题内容"]},
# {"key": "content", "values": ["正文内容"]}
# ]
# }
# }
async with aiohttp.ClientSession() as session:
async with session.post(url, headers=headers, json=payload) as resp:
return await resp.json()
Ark 模板消息(卡片消息)
# Ark 23:链接+文本列表
ark_23 = {
"ark": {
"template_id": 23,
"kv": [
{"key": "#DESC#", "value": "描述信息"},
{"key": "#PROMPT#", "value": "通知提醒"},
{"key": "#LIST#", "obj": [
{"obj_kv": [
{"key": "desc", "value": "项目1: 详情内容"}
]},
{"obj_kv": [
{"key": "desc", "value": "项目2: 详情内容"}
]}
]}
]
}
}
# Ark 24:文本+缩略图
ark_24 = {
"ark": {
"template_id": 24,
"kv": [
{"key": "#DESC#", "value": "描述"},
{"key": "#PROMPT#", "value": "提示"},
{"key": "#TITLE#", "value": "标题"},
{"key": "#METADESC#", "value": "详细描述"},
{"key": "#IMG#", "value": "https://example.com/image.png"},
{"key": "#LINK#", "value": "https://example.com"}
]
}
}
QQ 小程序
基础结构
QQ 小程序与微信小程序框架类似,使用 QXML + QSS + JS 开发:
├── app.js # 应用入口
├── app.json # 全局配置
├── app.qss # 全局样式
└── pages/
└── index/
├── index.js
├── index.json
├── index.qxml
└── index.qss
核心差异点:
- 模板文件后缀为
.qxml(对应微信.wxml) - 样式文件后缀为
.qss(对应微信.wxss) - API 前缀为
qq.xxx(对应微信wx.xxx) - 支持 QQ 特有能力:QQ 钱包支付、QQ 运动、厘米秀等
登录流程
// QQ 小程序登录
qq.login({
success(res) {
qq.request({
url: 'https://your-server.com/api/qq-login',
method: 'POST',
data: { code: res.code },
success(resp) {
qq.setStorageSync('token', resp.data.token);
}
});
}
});
# 后端:code 换取 session
def qq_miniapp_login(code):
url = (
'https://api.q.qq.com/sns/jscode2session'
f'?appid={QQ_APPID}&secret={QQ_SECRET}&js_code={code}'
'&grant_type=authorization_code'
)
resp = requests.get(url).json()
return resp # {"openid": "...", "session_key": "..."}
开放平台 OAuth2.0
授权流程
用户 → 授权页面 → 同意授权 → 回调获取 code → 换取 access_token → 获取用户信息
# 第一步:构造授权 URL
auth_url = (
"https://graph.qq.com/oauth2.0/authorize"
f"?response_type=code&client_id={APP_ID}"
f"&redirect_uri={REDIRECT_URI}&state={STATE}&scope=get_user_info"
)
# 第二步:code 换取 access_token
def get_qq_access_token(code: str) -> dict:
url = (
"https://graph.qq.com/oauth2.0/token"
f"?grant_type=authorization_code&client_id={APP_ID}"
f"&client_secret={APP_SECRET}&code={code}&redirect_uri={REDIRECT_URI}"
"&fmt=json"
)
return requests.get(url).json()
# {"access_token": "...", "expires_in": 7776000, "refresh_token": "..."}
# 第三步:获取 openid
def get_openid(access_token: str) -> str:
url = f"https://graph.qq.com/oauth2.0/me?access_token={access_token}&fmt=json"
data = requests.get(url).json()
return data["openid"]
# 第四步:获取用户信息
def get_user_info(access_token: str, openid: str) -> dict:
url = (
"https://graph.qq.com/user/get_user_info"
f"?access_token={access_token}&oauth_consumer_key={APP_ID}&openid={openid}"
)
return requests.get(url).json()
# {"nickname", "figureurl_qq", "gender", "province", "city", ...}
使用场景
- QQ 群机器人:关键词自动回复、群管理(禁言/踢人)、定时推送
- 频道内容管理:自动发帖、消息卡片推送、成员管理
- 客服机器人:C2C 私聊自动应答、FAQ 问答、工单转接
- 频道互动:表情表态统计、投票、签到打卡
- QQ 小程序:轻量级应用、游戏中心、社交分享
最佳实践
安全规范
- AppSecret 仅在服务端使用,禁止出现在前端代码
- access_token 需中心化缓存,避免频繁请求
- WebSocket 连接需处理断线重连(使用 session_id + seq 恢复)
- 群聊/C2C 消息必须被动回复(需携带 msg_id),不支持主动推送
频率限制
| API 类型 | 限制 |
|---|---|
| 全局 | 单机器人 800 次/分钟 |
| 频道消息 | 单子频道 5 条/秒 |
| 群消息 | 被动回复,无主动推送 |
| C2C 消息 | 被动回复,无主动推送 |
常用 API 端点
| 环境 | Base URL |
|---|---|
| 正式环境 | https://api.sgroup.qq.com |
| 沙箱环境 | https://sandbox.api.sgroup.qq.com |
| API | 方法 | 路径 |
|---|---|---|
| 获取网关 | GET | /gateway |
| 频道详情 | GET | /guilds/{guild_id} |
| 子频道列表 | GET | /guilds/{guild_id}/channels |
| 发送频道消息 | POST | /channels/{channel_id}/messages |
| 发送私信 | POST | /dms/{guild_id}/messages |
| 发送群消息 | POST | /v2/groups/{group_openid}/messages |
| 发送 C2C 消息 | POST | /v2/users/{openid}/messages |
| 频道成员列表 | GET | /guilds/{guild_id}/members |
| 禁言成员 | PATCH | /guilds/{guild_id}/members/{user_id}/mute |
断线重连
async def reconnect(self, ws_url: str):
"""使用 session_id 和 seq 恢复会话"""
async with aiohttp.ClientSession() as session:
async with session.ws_connect(ws_url) as ws:
# 发送 Resume(opcode=6)
await ws.send_json({
"op": 6,
"d": {
"token": f"QQBot {self.token}",
"session_id": self.session_id,
"seq": self.seq
}
})
# 恢复成功后继续接收事件
本 skill 不做什么
- 不提供 QQ 协议逆向、模拟客户端的实现细节(风控/封号高危, 社区路线仅指路 OneBot 标准与现役实现名字,风险自担)
- 官方群/C2C 机器人能力需在 q.qq.com 提审开通,未过审前接口会报 权限错误,属正常现象
- 接口参数以官方文档 bot.q.qq.com/wiki 为准,本文示例为骨架
最后更新: 2026-07-11
Questions people ask
- How does it help me choose between the official Bot API and OneBot?
- It compares compliance, supported scenarios, message restrictions, ecosystem status, and intended use. It recommends the official API for public or commercial products and limits OneBot guidance to personal use without protocol reverse-engineering details.
- Does it cover the official bot connection flow?
- Yes. It walks through token acquisition, gateway discovery, WebSocket Hello and Identify events, intent bitmasks, heartbeats, dispatching, and session recovery considerations.
- Which message and identity flows are included?
- It covers text, Markdown, Ark, Embed, and rich-media message types, including passive group and C2C replies with msg_id. It also documents QQ Mini Program code-to-session login and OAuth 2.0 authorization, token, openid, and user-info steps.
Related skills
Publish Official Account content by CLI and troubleshoot common WeChat platform integration issues.
Create WeChat Official Account article drafts and image-text posts from Markdown, HTML, plain text, and images.
Register an AI agent in OpenBotCity, configure secure access, and connect it to the city’s live manual.
Routes CloudBase work through scenario-specific implementation, deployment, troubleshooting, and review workflows.
Translate text, URLs, and files with selectable workflows, glossary control, and preserved Markdown structure.