← Back to Skills Marketplace
39
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install wechat-lead-generation
Description
微信潜在客户抓取、分析与自动回复营销自动化技能
README (SKILL.md)
WeChat Lead Generation 技能
自动抓取微信联系人/群聊数据,AI 分析潜在客户意向,智能生成个性化回复
核心功能
- 🔍 多渠道抓取 - 支持微信好友、群聊、朋友圈、公众号文章抓取
- 🧠 智能分析 - 基于对话内容识别客户兴趣、购买意向、客户画像
- 🤖 自动回复 - 根据客户画像生成个性化跟进话术
- 💾 线索存储 - 自动存入 agentmemory,支持长期追踪与评分
适用场景
- 销售团队自动跟进潜在客户
- 社群运营自动化互动
- 微商/电商客户转化
- 产品推广线索收集
输入参数
| 参数 | 类型 | 必填 | 说明 |
|---|---|---|---|
source |
string | ✅ | 数据来源:friends/groups/moments/articles |
days_back |
int | ⭕ | 抓取最近 N 天数据(默认 7) |
keywords |
array | ⭕ | 过滤关键词(如 "AI"、"机器人") |
analysis_depth |
enum | ⭕ | 分析深度:basic/detailed/deep |
auto_reply |
bool | ⭕ | 是否自动生成回复(默认 false) |
reply_template |
string | ⭕ | 回复模板(变量:{name}, {interest}, {product}) |
输出产物
- 线索报告 (
leads-report.md) - 潜在客户列表 + 评分 - 客户画像 (
artifacts/profiles.json) - 分析结果 JSON - 自动回复 (
artifacts/replies.md) - 待发送的回复草稿 - 记忆存储 - 客户数据存入向量数据库
工作流
graph LR
A[数据源: 微信好友/群聊] --> B[抓取对话/朋友圈]
B --> C[AI 分析意图 & 兴趣]
C --> D[客户画像 & 线索评分]
D --> E{auto_reply?}
E -->|是| F[生成个性化回复]
E -->|否| G[输出报告供人工审核]
F --> H[存入 agentmemory]
G --> H
使用示例
抓取并分析群聊潜在客户
# 基础分析(不自动回复)
wechat-lead-generation --source groups --days_back 3 --keywords "AI,chatbot" --analysis_depth detailed
# 完整流程(自动生成回复)
wechat-lead-generation \
--source friends \
--days_back 7 \
--analysis_depth deep \
--auto_reply true \
--reply_template "你好{name},注意到你对{interest}感兴趣,我们的{product}可能适合你,..."
定时任务
每天早上 9:00 自动扫描新线索:
openclaw cron add \
--name "微信线索自动扫描" \
--schedule "0 9 * * *" \
--payload.agentTurn.message "用 wechat-lead-generation 抓取今天群聊中的 AI 相关线索,分析深度 detailed,生成报告但不自动回复" \
--delivery.announce.channel openclaw-weixin
配置依赖
必需技能
wechat-md-publish或bb-browser-openclaw- 微信数据抓取trendradar- 行业热点关联(可选)agentmemory- 客户画像存储
环境变量
# 如果使用 wechat-md-publish 需要配置微信 cookie
WECHAT_COOKIE="your_wechat_cookie_here"
# 如需自动回复(谨慎使用,避免封号)
AUTO_REPLY_ENABLED=true
风险与合规
⚠️ 重要提示:
- 微信自动抓取和自动回复可能违反微信用户协议
- 建议使用 半自动模式(生成回复草稿供人工审核)
- 避免高频操作(建议间隔 > 30 秒)
- 仅用于合法合规的客户跟进场景
- 用户需自行承担使用风险
输出结构
output/wechat-lead-generation/
├── leads-report-20260526.md # 线索汇总报告
├── artifacts/
│ ├── profiles.json # 客户画像数据
│ ├── high_score_leads.json # 高评分线索(>80分)
│ ├── replies.md # 自动回复草稿
│ └── raw_messages.json # 原始抓取数据
评分算法
线索评分(0-100)基于:
| 维度 | 权重 | 说明 |
|---|---|---|
| 关键词匹配 | 30% | 对话中出现产品相关关键词 |
| 互动频率 | 25% | 与你的消息互动次数 |
| 兴趣强度 | 25% | 主动询问、表达购买意向 |
| 最近联系 | 20% | 时间越近分数越高 |
Troubleshooting
抓取失败
- 检查微信 cookie 是否有效
- 确认
wechat-md-publish已安装并配置 - 降低抓取频率,避免触发风控
生成回复质量低
- 增加
--analysis_depth deep - 提供更多上下文数据(更多天历史)
- 自定义
--reply_template包含个性化字段
被封号风险
- 启用
--auto_reply false转为人工审核模式 - 设置 cron 任务间隔至少 4 小时一次
- 使用真实微信号(非小号)并保持合理频率
License
MIT © Chace
Usage Guidance
Only install this if you are comfortable with a tool that may process WeChat conversations, contacts, and lead profiles for marketing use. Treat it as requiring explicit consent and a privacy review before real WeChat data is connected; keep auto-reply off, avoid scheduled scraping, and delete or disable raw message/profile storage unless you have a clear lawful basis and retention plan.
Capability Assessment
Purpose & Capability
The stated purpose, WeChat lead generation and automated follow-up, matches the major capabilities, but the capability itself involves scraping private/social communications, profiling people, and generating outreach drafts.
Instruction Scope
The workflow discloses some risk and recommends semi-automatic use, but it also encourages scheduled scans and long-term lead tracking without clear data minimization, consent gates, or deletion controls.
Install Mechanism
The install instructions are ordinary local skill setup steps, but metadata declares no capability tags despite documented WeChat credential/session use, file writes, memory storage, and cron-style operation.
Credentials
Access to WeChat cookies or login state, private messages, contacts/groups, and local persistent artifacts is high impact for a marketing automation skill and is not tightly scoped.
Persistence & Privilege
The engine writes reports, profiles, high-score leads, raw messages, and optionally agentmemory records containing names and message excerpts, with no visible retention or redaction policy.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install wechat-lead-generation - After installation, invoke the skill by name or use
/wechat-lead-generation - Provide required inputs per the skill's parameter spec and get structured output
Version History
v0.1.0
wechat-lead-generation 0.1.0 — Initial Release
- Automates WeChat lead capture, intent analysis, and personalized reply generation.
- Supports multi-source data collection: friends, groups, Moments, and articles.
- AI analyzes conversations for customer interests, scoring potential leads.
- Integrates with agentmemory for long-term profile storage and lead tracking.
- Outputs include detailed lead reports, customer profiles, and reply drafts.
- Offers batch/cron operation, customizable analysis depth, and reply templates.
Metadata
Frequently Asked Questions
What is WeChat Lead Generation?
微信潜在客户抓取、分析与自动回复营销自动化技能. It is an AI Agent Skill for Claude Code / OpenClaw, with 39 downloads so far.
How do I install WeChat Lead Generation?
Run "/install wechat-lead-generation" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is WeChat Lead Generation free?
Yes, WeChat Lead Generation is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does WeChat Lead Generation support?
WeChat Lead Generation is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created WeChat Lead Generation?
It is built and maintained by Chace (@ling-qian); the current version is v0.1.0.
More Skills