mcp-timely
A read-only MCP server for querying Timely time tracking data, providing tools for project overviews, time spent summaries, and work log entries.
README
mcp-timely
MCP server for Timely time tracking — built for asking questions, not wrapping endpoints. Three read-only tools, each answered in a single upstream call with Timely doing the aggregation server-side, so an LLM never pages through raw entries to sum hours (and never gets the sum wrong).
| Tool | The question it answers |
|---|---|
projects_overview |
What's on my plate? — projects with client, logged hours, budget burn, unbilled amounts |
time_spent |
Where did my time go? — date-range rollups grouped by project, client, label, or day, with billable split |
work_log |
What did I actually do? — individual entries with notes for a range; standup/diary/invoicing material |
Scope is deliberate: read-only, and always scoped to the authorized user. There are no write tools and no user/team parameters.
Setup
-
Create a Timely OAuth app at
https://app.timelyapp.com/<account>/oauth_applicationswith callback URLurn:ietf:wg:oauth:2.0:oob. -
Configure
cp .env.example .env # fill in TIMELY_CLIENT_ID and TIMELY_CLIENT_SECRET -
Authorize (once)
uv sync uv run mcp-timely authOpen the printed URL, authorize, paste the code. Tokens land in
TIMELY_TOKEN_FILE(defaulttimely_tokens.json) and refresh themselves from then on. -
Add to Claude Code
claude mcp add timely -- uv run --directory /path/to/mcp-timely mcp-timely
Token handling
Timely rotates refresh tokens on every refresh and its access tokens carry no expiry. The session layer uses the access token until a 401, then refreshes once — persisting the new pair atomically before continuing — and retries the request once. The token file is the only state; keep it on a persistent volume in Docker. Re-authorization is only needed if the grant is revoked.
HTTP deployment
Set TRANSPORT=http and MCP_API_KEY (the server refuses to start
unauthenticated HTTP). Serves streamable-http on /mcp, liveness on
/health. See compose.yaml.
Development
uv run pytest # unit tests (in-memory MCP client, no network)
uv run ruff check .
Releases
Every push to main that touches non-doc files releases automatically: tests
and a security audit run, the patch version bumps, a vX.Y.Z tag lands, and a
multi-arch image is published to ghcr.io/caseyro/mcp-timely. Consequences:
- Don't edit
versioninpyproject.tomlby hand — CI owns it. - Add
[skip ci]to a commit message to skip a release. - Rebase PRs on
mainbefore merging so the auto-bump commit doesn't race yours. - Markdown-only and test-only changes don't trigger a release.
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。