keka-mcp-server
MCP server for the Keka HRM API, enabling HR operations like employee management, leave, attendance, payroll, recruitment, and PSA.
README
keka-mcp-server
MCP server for the Keka HRM API, built for White Label IQ (WLIQ).
Exposes 14 tools covering Keka's core modules so Claude (or any MCP client) can query and act on HR data.
Tools
HRIS
| Tool | Description |
|---|---|
keka_list_employees |
List employees with filters (status, search, probation, etc.) |
keka_get_employee |
Get full details for a single employee by ID |
keka_list_departments |
List all departments |
keka_list_job_titles |
List all job titles |
keka_list_groups |
List all groups/teams |
Leave Management
| Tool | Description |
|---|---|
keka_list_leave_types |
List configured leave types (Annual, Sick, etc.) |
keka_list_leave_requests |
List leave requests with date/employee filters |
keka_create_leave_request |
Submit a new leave request for an employee |
keka_get_leave_balances |
Get leave balance breakdown per employee |
Attendance
| Tool | Description |
|---|---|
keka_get_attendance |
Get attendance records (clock-in/out, hours, status) |
Payroll
| Tool | Description |
|---|---|
keka_list_pay_groups |
List payroll groups |
keka_list_pay_bands |
List salary pay bands |
keka_list_salaries |
List employee salaries (CTC, pay group) |
Recruitment (Keka Hire)
| Tool | Description |
|---|---|
keka_list_jobs |
List job openings |
keka_list_candidates |
List candidates for a specific job |
PSA
| Tool | Description |
|---|---|
keka_list_psa_clients |
List PSA clients |
keka_list_psa_projects |
List PSA projects |
Setup
1. Get Keka API credentials
In your Keka admin portal:
- Go to Settings → Integrations → API
- Create a new API access key
- Note down:
Client ID,Client Secret,API Key
2. Install dependencies
npm install
3. Build
npm run build
4. Environment variables
| Variable | Required | Description |
|---|---|---|
KEKA_BASE_URL |
✅ | Your Keka tenant URL, e.g. https://yourcompany.keka.com |
KEKA_CLIENT_ID |
✅ | OAuth2 Client ID from Keka admin |
KEKA_CLIENT_SECRET |
✅ | OAuth2 Client Secret |
KEKA_API_KEY |
✅ | API key from Keka admin |
KEKA_EMPLOYEE_ID |
Optional | Your own Keka employee ID — used as requestedBy when applying leave on behalf of others |
KEKA_SANDBOX |
Optional | Set to true to use kekademo.com sandbox auth |
TRANSPORT |
Optional | stdio (default) or http |
PORT |
Optional | HTTP port when TRANSPORT=http (default: 3000) |
Claude Desktop integration (stdio)
Add to your claude_desktop_config.json:
{
"mcpServers": {
"keka": {
"command": "node",
"args": ["/path/to/keka-mcp-server/dist/index.js"],
"env": {
"KEKA_BASE_URL": "https://yourcompany.keka.com",
"KEKA_CLIENT_ID": "your-client-id",
"KEKA_CLIENT_SECRET": "your-client-secret",
"KEKA_API_KEY": "your-api-key",
"KEKA_EMPLOYEE_ID": "your-employee-uuid"
}
}
}
}
HR Skill Prompt (optional)
Copy the contents of keka-hr-skill.md into Claude Desktop's Custom Instructions (Settings → Custom Instructions). This teaches Claude how to handle Keka operations smoothly — resolving names to IDs, always including required fields, checking balances before applying leave, etc.
## HTTP mode (remote deployment)
```bash
TRANSPORT=http \
KEKA_BASE_URL=https://yourcompany.keka.com \
KEKA_CLIENT_ID=xxx \
KEKA_CLIENT_SECRET=xxx \
KEKA_API_KEY=xxx \
node dist/index.js
Health check: GET http://localhost:3000/health
MCP endpoint: POST http://localhost:3000/mcp
Authentication
Keka uses a custom OAuth2 flow (grant_type=kekaapi). The server:
- Fetches a Bearer token on first request
- Caches the token in memory
- Auto-refreshes 2 minutes before expiry (tokens last 24 hours)
Rate Limits
Keka enforces 50 requests per minute. The server returns a clear error message if this limit is hit.
Development
# Watch mode (no build required)
npm run dev
# Build
npm run build
# Start production
npm start
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。
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。
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。
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。
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。
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。
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
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。