HR Assist MCP Server
Agentic HR assistant providing employee, leave, meeting, ticket, and email tools over the Model Context Protocol, designed to automate onboarding workflows via an MCP client such as Claude Desktop.
README
HR Assist (MCP Server)
Agentic HR assistant that exposes employee, leave, meeting, ticket, and email tools over the Model Context Protocol (MCP). Designed to automate onboarding workflows via an MCP client such as Claude Desktop.
Sample employees & managers
Use these when creating a new employee (manager_id can be ID or name):
| ID | Name | Role | Department |
|---|---|---|---|
| E001 | Sarah Johnson | VP Engineering | Engineering |
| E002 | Michael Chen | VP Product | Product |
| E003 | David Wilson | Engineering Manager | Engineering |
| E006 | Emily Kim | Product Manager | Product |
| E009 | Priya Nair | HR Manager | HR |
Other sample staff: Tony Sharma (E004), James Rodriguez (E005), Carlos Mendez (E007), Lisa Wong (E008), Aisha Khan (E010), Rohan Mehta (E011), Neha Kapoor (E012).
Examples:
add_employee("Riya Shah", manager_id="E001", email="riya.shah@atliq.com")
add_employee("Riya Shah", manager_id="Sarah Johnson")
add_employee("Riya Shah") # defaults manager to E001
Call list_managers or list_employees tools to see the full directory.
Features
- Add / look up employees
- Leave balance, apply leave, leave history
- Schedule / list / cancel meetings
- Create and manage IT/HR tickets
- Optional Gmail sending for welcome / manager notifications
- Built-in onboarding prompt:
onboard_new_employee
Setup
-
Install uv (recommended) or use pip.
-
From this directory:
uv syncOr with pip:
python -m venv .venv .venv\Scripts\activate pip install -e . -
Copy env values:
copy sample.env .envOptional (for real email):
CB_EMAIL=your@gmail.com CB_EMAIL_PWD=your_gmail_app_passwordLeave blank to run without email (send_email will skip and report what it would send).
Run Streamlit web app (recommended UI)
uv sync
uv run streamlit run streamlit_app.py
Opens at http://localhost:8501 with tabs for Directory, Onboard, Leave, Tickets, Meetings, Email, and Lookup.
Email setup (Send email tab)
Local .env:
CB_EMAIL=yourgmail@gmail.com
CB_EMAIL_PWD=your_16_char_app_password
Streamlit Cloud → Settings → Secrets:
CB_EMAIL = "yourgmail@gmail.com"
CB_EMAIL_PWD = "your_16_char_app_password"
Use a Gmail App Password (not your normal password).
Deploy to Streamlit Community Cloud
- Push this folder to a GitHub repo (include
streamlit_app.pyandrequirements.txt). - Go to share.streamlit.io → New app.
- Select the repo, set Main file path to
streamlit_app.py. - Deploy. You’ll get a public URL like
https://your-app.streamlit.app.
Run locally (CLI demo, no UI)
uv run python run_local.py
This seeds dummy HR data and runs an onboarding-style flow through the same managers used by the MCP tools.
Run MCP server
uv run python server.py
stdio transport waits for an MCP client. To inspect tools interactively:
npx @modelcontextprotocol/inspector uv run python server.py
Claude Desktop config example
Edit Claude Desktop claude_desktop_config.json and point command / --directory at this project folder:
{
"mcpServers": {
"hr-assist": {
"command": "uv",
"args": [
"--directory",
"C:\\Users\\omkar\\Documents\\Project\\AI\\HR Assist (MCP Server)",
"run",
"python",
"server.py"
],
"env": {
"CB_EMAIL": "YOUR_EMAIL",
"CB_EMAIL_PWD": "YOUR_APP_PASSWORD"
}
}
}
}
Usage
In Claude Desktop, use the onboard_new_employee prompt or ask naturally, for example:
- Onboard a new employee named Riya reporting to Sarah Johnson
- What is Tony Sharma's leave balance?
- Create a laptop ticket for E009
Project layout
server.py MCP tools + onboarding prompt
run_local.py Local demo without MCP client
emails.py SMTP helper
utils.py Dummy data seeder
hrms/ Employee, leave, meeting, ticket managers + schemas
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
mcp-server-qdrant
这个仓库展示了如何为向量搜索引擎 Qdrant 创建一个 MCP (Managed Control Plane) 服务器的示例。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器