clockify-mcp
Read-only MCP server for Clockify time tracking, exposing 60 raw read tools and five workflows for status, workspace overview, and review of days/weeks.
README
clockify-python-115
clockify-python-115 is an independent community project. It is not affiliated
with, endorsed by, or sponsored by CAKE.com or Clockify. The project uses the
Clockify name only to identify the service that it supports.
The distribution contains:
clockify: a typed async SDK with all 168 known Clockify operations on 29 resources.clockify_mcp: a structurally read-only Model Context Protocol (MCP) server.
Release status: 0.1.0 is a local release candidate. It is not published to
PyPI yet. The default MCP server registers 60 raw reads and five workflows. It
registers zero writes.
Install
Install the local release candidate after uv build:
uv pip install dist/clockify_python_115-0.1.0-py3-none-any.whl
uv pip install "dist/clockify_python_115-0.1.0-py3-none-any.whl[mcp]"
After publication, use:
uv add clockify-python-115
uv add "clockify-python-115[mcp]"
Python 3.11, 3.12, 3.13, and 3.14 are supported and tested.
Read-only SDK quickstart
import asyncio
import os
from clockify import ClockifyClient
async def main() -> None:
async with ClockifyClient(
api_key=os.environ["CLOCKIFY_API_KEY"],
workspace_id=os.environ.get("CLOCKIFY_WORKSPACE_ID"),
) as clockify:
me = await clockify.users.me()
projects = await clockify.projects.list(archived=False, page_size=25)
print(me.name, len(projects))
asyncio.run(main())
See docs/quickstart.md and examples/sdk_list_projects.py.
MCP quickstart
clockify-mcp
Configure exactly one of CLOCKIFY_API_KEY or CLOCKIFY_ADDON_TOKEN. You can
also configure CLOCKIFY_WORKSPACE_ID.
The MCP contract is exact:
- 60 raw read tools;
- five workflows:
clockify_status,clockify_workspace_overview,clockify_review_day,clockify_review_week, andclockify_doctor; - 65 tools in total;
- zero registered writes.
The SDK exposes writes to explicit Python callers. MCP does not. See docs/mcp-guide.md and examples/mcp_config.example.json.
SDK behavior
- Configure exactly one credential. The SDK rejects caller-supplied authority and Clockify credential headers before network access.
- Ordinary caller headers and
X-Request-Idare preserved. - Read retries support delay-seconds and HTTP-date
Retry-Aftervalues. - No write is automatically retried. An ambiguous write transport failure
raises
MutationOutcomeUnknownError. Read state before a manual retry. - Pagination names vary by operation. Use
iter_pagesoriter_alland honor the operation'sLast-Pagebehavior. - Reports and audit-log operations use their required service hosts automatically.
- Some
PUToperations replace the complete entity. Resend each field that must remain. See docs/api-deviations.md. - API errors expose bounded sanitized detail, status, operation ID, request ID, safe API code, and safe retry timing.
The direct write example is intentionally separate. It mutates only a verified sacrificial workspace and uses a unique name: examples/sdk_create_tag_sacrificial.py.
Develop
uv sync --all-extras --dev
uv run ruff check .
uv run ruff format --check .
uv run pyright
uv run pytest -q -m "not live"
uv build
See CONTRIBUTING.md, SECURITY.md, NOTICE.md, and LICENSE.
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。