openhr

$npx mdskill add aAAaqwq/AGI-Super-Team/openhr

Automate full Boss recruitment with login, greeting, chat, and resume parsing.

  • Executes automated candidate screening and personalized outreach.
  • Integrates with Boss Zhipin, Feishu, and headed browser automation.
  • Uses LLM for intelligent chat responses and resume data extraction.
  • Delivers status updates and progress tracking to the user.

SKILL.md

.github/skills/openhrView on GitHub ↗
---
name: openhr
description: Boss直聘全流程AI自动化招聘智能体。自动登录、筛选候选人、个性化打招呼、智能聊天跟进、简历解析、飞书同步。触发词：启动招聘、自动打招呼、boss greet、openhr、简历解析、上传飞书。
license: MIT
metadata:
  author: 小code
  version: "1.0"
read_when:
  - User mentions "openhr" or "招聘" or "boss" or "打招呼"
  - User says "打招呼" or "启动招聘" in recruitment context
  - User wants to start automated recruiting or check recruiting progress
  - User wants to parse resumes or upload to Feishu
triggers:
  - openhr
  - openhr start
  - openhr greet
  - openhr candidates
  - openhr chat
  - openhr monitor
  - openhr status
  - openhr resume
  - openhr feishu
  - boss greet
  - boss login
  - 启动招聘
  - 自动打招呼
  - 打招呼
  - 查看候选人
  - 启动聊天引擎
  - 招聘监控
  - 登录Boss
  - 查看招聘进度
  - 简历解析
  - 上传飞书
---

# OpenHR — Boss直聘自动化招聘智能体

> 6 大模块全部通过验收 ✅
> 项目路径：`~/clawd/projects/openhr/`

---

## ⚠️ 安全警告（必读）

1. **Boss 直聘有严格反爬虫机制**，所有操作间隔必须 ≥ 5 秒
2. **每日打招呼上限 100 次**（Boss 官方产品限制，不可突破）
3. **验证码出现时必须暂停**，等人工处理完成后再继续
4. **封号风险提示**：频繁操作、异常行为可能触发风控导致封号
5. **必须使用 headed 浏览器**（有界面，会弹窗），不能 headless
6. **首次使用需要人工扫码登录**
7. **建议先用 `--dry-run` 测试，确认无误后再正式运行**

---

## 路由表

| 触发词 | 子模块 | 执行脚本 | 说明 |
|--------|--------|----------|------|
| 登录Boss / boss login | boss-login | `scripts/boss_login.py` | 扫码登录 + Cookie 持久化 |
| 自动打招呼 / openhr greet / boss greet / 打招呼 | boss-greet | `scripts/boss_greet.py` | 遍历筛选 + 个性化打招呼 |
| 启动聊天引擎 / openhr chat | chat-engine | `scripts/chat_engine.py` | 监控回复 + LLM 智能跟进 |
| 简历解析 / openhr resume | resume-parser | `scripts/resume_parser.py` | 提取简历信息 |
| 上传飞书 / openhr feishu | resume-parser | `scripts/feishu_upload.py` | 写入飞书多维表格 |
| 查看候选人 / openhr candidates | boss-greet | `scripts/boss_greet.py --list` | 列出已筛选候选人 |
| 招聘监控 / openhr monitor | chat-engine | `scripts/chat_engine.py --monitor` | 持续监控聊天消息 |
| 查看招聘进度 / openhr status | — | 直接读取状态 | 汇总 greet_count.json + chat_logs |
| 启动招聘 | boss-greet | `scripts/boss_greet.py` | 完整招聘流程入口 |

---

## 模块一览

| 模块 | 脚本 | 用途 |
|------|------|------|
| 登录 | `scripts/boss_login.py` | 扫码登录 + Cookie 持久化 |
| 风控 | `scripts/anti_detect.py` | 随机间隔 + 验证码检测 + 封号预警 |
| 打招呼 | `scripts/boss_greet.py` | 自动筛选 + 个性化打招呼 |
| 聊天 | `scripts/chat_engine.py` | 监控回复 + LLM 智能跟进 |
| 简历 | `scripts/resume_parser.py` | 简历信息提取 |
| 飞书 | `scripts/feishu_upload.py` | 飞书多维表格写入 |
| 知识库 | `scripts/knowledge_base.py` | 话术 + 岗位需求管理 |

---

## 快速开始

### 第一步：登录 Boss 直聘

```bash
cd ~/clawd/projects/openhr
python scripts/boss_login.py
```

- 首次运行会弹出浏览器二维码
- 用 Boss 直聘 APP 扫码确认登录
- Cookie 自动持久化到 `data/cookies/boss_cookies.json`
- 后续运行无需重复登录（Cookie 过期需重新扫码）

### 第二步：Dry-run 测试（强烈建议）

```bash
cd ~/clawd/projects/openhr
python scripts/boss_greet.py --dry-run
```

- 只筛选候选人，**不发送任何消息**
- 预览打招呼话术，确认无误后再正式运行
- 查看 `data/logs/dry_run.json` 了解模拟结果

### 第三步：正式打招呼

```bash
cd ~/clawd/projects/openhr
python scripts/boss_greet.py
```

- 默认每日上限 100 次（自动遵守 Boss 限制）
- 所有操作间隔 5-15 秒随机（内置风控保护）
- 运行日志实时输出，可随时 Ctrl+C 中断

### 第四步：启动聊天引擎

```bash
cd ~/clawd/projects/openhr
python scripts/chat_engine.py
```

- 监控新消息，LLM 生成智能回复
- 支持多种聊天模式（主动跟进、被动响应、催入职等）
- 聊天记录存入 `data/chat_logs/`

---

## 查看招聘进度

```bash
# 直接读取状态文件
cat ~/clawd/projects/openhr/data/greet_count.json

# 或运行 status 命令
cd ~/clawd/projects/openhr
python -c "
import json, os
gc = json.load(open('data/greet_count.json'))
print(f'打招呼总数: {gc.get(\"total\",0)}')
print(f'今日打招呼: {gc.get(\"today\",0)}')
print(f'昨日打招呼: {gc.get(\"yesterday\",0)}')
logs = sorted([f for f in os.listdir('data/chat_logs') if f.endswith('.json')])
print(f'聊天记录数: {len(logs)}')
"
```

---

## 简历解析 + 飞书同步

```bash
# 解析简历（支持 PDF、DOCX、图片）
cd ~/clawd/projects/openhr
python scripts/resume_parser.py <简历文件路径>

# 上传解析结果到飞书多维表格
python scripts/feishu_upload.py <简历文件路径>
```

飞书配置：`config/feishu.json`（需填写 app_id、table_id）

---

## 配置文件

| 文件 | 说明 |
|------|------|
| `config/anti_detect.json` | 风控参数（间隔、验证码检测规则） |
| `config/feishu.json` | 飞书多维表格配置（app_id、table_id、fields） |
| `config/templates.json` | 打招呼话术模板（8 个场景） |
| `config/llm.json` | LLM 模型配置（provider、model、api_key） |
| `data/knowledge/` | 岗位需求、话术库、聊天模式、反馈模板 |

---

## 子模块详细文档

- [boss-login](file://~/clawd/projects/openhr/skills/boss-login/SKILL.md)
- [anti-detect](file://~/clawd/projects/openhr/skills/anti-detect/SKILL.md)
- [boss-greet](file://~/clawd/projects/openhr/skills/boss-greet/SKILL.md)
- [chat-engine](file://~/clawd/projects/openhr/skills/chat-engine/SKILL.md)
- [resume-parser](file://~/clawd/projects/openhr/skills/resume-parser/SKILL.md)
- [knowledge-base](file://~/clawd/projects/openhr/skills/knowledge-base/SKILL.md)

---

## 执行命令速查

```bash
cd ~/clawd/projects/openhr

# 登录
python scripts/boss_login.py

# 打招呼（测试）
python scripts/boss_greet.py --dry-run

# 打招呼（正式）
python scripts/boss_greet.py

# 列出候选人
python scripts/boss_greet.py --list

# 聊天引擎
python scripts/chat_engine.py

# 聊天监控
python scripts/chat_engine.py --monitor

# 简历解析
python scripts/resume_parser.py path/to/resume.pdf

# 飞书上传统一简历
python scripts/feishu_upload.py path/to/resume.pdf
```

More from aAAaqwq/AGI-Super-Team

SkillDescription
a-fund-monitor监控 A 股基金实时估值与盘后净值，自动判断交易日并生成提醒或分析。
add-leadAdd company/person/relationship to CRM
ads-agentAI-агент для управления Facebook рекламой. Вызывай для анализа, оптимизации, создания кампаний и отчётов.
afrexai-compliance-auditRun internal compliance audits against major governance and security
afrexai-personal-financeComplete personal finance system — budgeting, debt payoff, investing, tax optimization, net worth tracking, and financial independence planning. Use when managing money, building wealth, paying off debt, planning retirement, or optimizing taxes. Zero dependencies.
after-salesUse when managing post-purchase experience, building customer loyalty, or increasing repeat purchases
agent-contactsAI agent contacts — add, list, remove MCP contacts. Use when someone gives an agent URL, or when you need to view/remove contacts.
agent-model-switcher批量查看和切换子 agent 的模型配置，用于统一调整多 agent 的 provider/model 设置。
agent-patternsFormat SPAWN REQUEST messages to launch parallel agents, generate structured agent status reports, and define communication protocols within the sprint system. Use when the user needs to coordinate multiple agents, format spawn requests, produce agent reports, or establish inter-agent communication patterns.
agent-task-confirmUse when confirming whether a dispatched agent task was actually received, activated, and progressing after sessions_send or other task handoff actions.