@lewinnovation/clockify-mcp-server
MCP server that enables AI agents to interact with Clockify time tracking via curated workflows and a generic API tool for managing workspaces, projects, tasks, and time entries.
README
@lewinnovation/clockify-mcp-server
MCP (Model Context Protocol) server for the Clockify REST API. It exposes a small set of curated tools for common workflows plus a generic clockify_request tool for other documented /v1/... endpoints.
Requirements
- Node.js 20+
- pnpm 10+
- A Clockify API key (profile settings)
Install
# via npx
npx @lewinnovation/clockify-mcp-server
# OR via pnpm globally
pnpm add -g @lewinnovation/clockify-mcp-server
# OR clone and install locally
pnpm install
pnpm run build
Configuration
| Variable | Required | Description |
|---|---|---|
CLOCKIFY_API_KEY |
Yes | Sent as X-Api-Key on every request. |
CLOCKIFY_API_BASE_URL |
No | Defaults to https://api.clockify.me/api. Override only if Clockify documents a different base for your account or region. |
Run (stdio)
The server speaks MCP over stdio (stdout must stay clean JSON-RPC; logs go to stderr).
export CLOCKIFY_API_KEY="your-key"
pnpm start
Or after pnpm run build:
node dist/index.js
MCP tool handlers are registered from src/tools/register.ts (shared helpers in src/tools/helpers.ts).
MCP client setup (popular AI agent tools)
Use this server definition in your MCP client:
{
"mcpServers": {
"clockify": {
"command": "npx",
"args": ["-y", "@lewinnovation/clockify-mcp-server"],
"env": {
"CLOCKIFY_API_KEY": "your-key"
}
}
}
}
If you prefer running a local clone, use:
{
"mcpServers": {
"clockify": {
"command": "node",
"args": ["/absolute/path/to/clockify-mcp-server/dist/index.js"],
"env": {
"CLOCKIFY_API_KEY": "your-key"
}
}
}
}
Cursor
Add the JSON entry above to:
- Workspace config:
.cursor/mcp.json - Or user config:
~/.cursor/mcp.json
Then reload Cursor (or restart the MCP server from Cursor settings).
Claude Desktop
Add the same mcpServers.clockify entry to Claude Desktop MCP config:
- macOS:
~/Library/Application Support/Claude/claude_desktop_config.json
Restart Claude Desktop after saving the config.
Claude Code (CLI)
From your project directory:
claude mcp add clockify --env CLOCKIFY_API_KEY=your-key -- npx -y @lewinnovation/clockify-mcp-server
Or add the same JSON server entry to your Claude Code MCP config file if you manage MCP servers declaratively.
Other MCP-compatible agent tools
For tools such as Cline, Roo Code, and Windsurf, add the same server command/env block in that tool's MCP server settings UI or config file:
command:npxargs:["-y", "@lewinnovation/clockify-mcp-server"]env.CLOCKIFY_API_KEY: your Clockify API key
The server uses stdio transport, so no host/port configuration is required.
Tools
Curated
get_current_user—GET /v1/userlist_workspaces—GET /v1/workspaces(optional singlerolefilter; multiple roles requireclockify_request)list_projects—GET /v1/workspaces/{workspaceId}/projectswith pagination (page,page_size→page-size)list_tasks—GET /v1/workspaces/{workspaceId}/projects/{projectId}/taskslist_time_entries—GET /v1/workspaces/{workspaceId}/user/{userId}/time-entries(ifuser_idis omitted, the current API user is resolved viaGET /v1/user)create_time_entry—POST /v1/workspaces/{workspaceId}/time-entriesupdate_time_entry—PUT /v1/workspaces/{workspaceId}/time-entries/{id}(Clockify requiresstarton updates)stop_running_time_entry—PATCH /v1/workspaces/{workspaceId}/user/{userId}/time-entrieswith{ end }(optionaluser_id, default current user)
Generic
clockify_request—method,path(must start with/v1/), optionalquerymap, optional JSONbody(max 256 KiB serialized). Array query values are sent as repeated keys.
Responses wrap JSON in a text content block; when Clockify returns a Last-Page header, curated list-style tools include lastPage in the decoded JSON payload.
OpenAPI types
Types are generated from Clockify’s OpenAPI document:
pnpm run generate:openapi
Source URL: https://api.clockify.me/api/v3/api-docs
Output: src/generated/clockify.d.ts
Re-run this periodically to pick up API changes, then fix any compile errors from type drift.
Scripts
| Script | Purpose |
|---|---|
pnpm run dev |
Watch src/index.ts with tsx. |
pnpm run build |
Emit dist/ with tsc. |
pnpm run typecheck |
tsc --noEmit. |
pnpm run smoke |
Mocked HTTP smoke test for curated endpoints (no API key needed; uses a dummy key + mocked fetch). |
pnpm run generate:openapi |
Regenerate src/generated/clockify.d.ts. |
License
MIT
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。