upservice-mcp

upservice-mcp

Manages employees, projects, sprints, tasks, tags, directories, and channel messages via the Upservice Public API.

Category
访问服务器

README

Upservice MCP Server

Читать на русском

An MCP (Model Context Protocol) server that wraps the Upservice Public API so Claude (or any MCP client) can manage employees, projects, sprints, tasks, tags, directories, and channel messages directly.

Built from the OpenAPI spec at https://public.upservice.io/openapi.json, verified against the live API.

What it can do

58 tools covering:

  • Employees — list employees
  • Projects — list, create, get, update, delete, set managers/members, mark completed
  • Sprints — list, create, get, update, delete, complete, activate, add tasks
  • Tags — list, create, update, delete, assign/unassign to tasks/chats/assets/contacts/attachments
  • Tasks — list/search, create, get, update, delete, attachments, status changes, effort estimation, worklog, agreement (approval) workflow actions, co-responsibles, agreement/acquaintance sheets
  • Directories (custom reference catalogs) — list/create/get/update/delete directories and their records, plus bulk relation management (linking records to tasks, projects, orders, etc.)
  • Channels & files — load/send chat messages, send channel messages, upload files, get file download URLs

Every write tool exposes the well-documented fields explicitly (validated), and most also accept an extra_fields dict for advanced/uncommon fields documented in the Upservice API but not modeled individually.

Mentioning employees: Upservice only turns @Name into a real, notifying mention if it's written as @[Name](employee_id). Tools that write text (upservice_send_chat_message, upservice_send_channel_message, upservice_create_task, upservice_update_task) accept a mentions: [{employee_id, display_name}] field — put a {{employee_id}} placeholder in your text and the server substitutes the correct syntax for you.

Known API limitation: GET /v1/tasks (used by upservice_list_tasks) has no status/is_completed filter — confirmed against the live API, not just the docs. Filter by status client-side after narrowing with date_end_gte/date_end_lte, project, author, or responsible.

Setup

Get your Upservice API key first: in your Upservice account, go to account settings → API key. Keys look like UPS-XXXX-XXXX-XXXX-XXXX.

Option A — uvx (recommended, no manual install)

Requires uv installed (brew install uv or see their docs). Add to your MCP client config (e.g. claude_desktop_config.json):

{
  "mcpServers": {
    "upservice": {
      "command": "uvx",
      "args": ["--from", "/absolute/path/to/upservice_mcp", "upservice-mcp"],
      "env": {
        "UPSERVICE_API_KEY": "UPS-XXXX-XXXX-XXXX-XXXX"
      }
    }
  }
}

If this project is published to a git remote, --from can point at it directly instead of a local path: "args": ["--from", "git+https://github.com/alexherbaly/upservice-mcp", "upservice-mcp"] — then a colleague only needs the URL and their own API key, no file copying at all.

Pin to a ref rather than the bare URL — without one, uvx tracks the main branch, so a bad push to main breaks everyone's server on their next run. Two options:

  • @stable (recommended for colleagues) — a tag that CI automatically moves to the tip of main every time the build passes. No one has to remember to cut a release for routine fixes; everyone's server just picks up the latest known-good commit next time it restarts.
  • @v0.2.0 (a specific version) — frozen forever at that exact commit, for when you want a name you can point back to later (release notes, "the version we tested on date X"). Cut these manually with git tag when it's worth a name, not on every push.

Example: "args": ["--from", "git+https://github.com/alexherbaly/upservice-mcp@stable", "upservice-mcp"]

uv builds and caches an isolated environment on first run; nothing is installed system-wide.

Option B — plain venv + pip

cd upservice_mcp
python3 -m venv venv
venv/bin/pip install -r requirements.txt
{
  "mcpServers": {
    "upservice": {
      "command": "/absolute/path/to/upservice_mcp/venv/bin/python3",
      "args": ["/absolute/path/to/upservice_mcp/src/upservice_mcp/server.py"],
      "env": {
        "UPSERVICE_API_KEY": "UPS-XXXX-XXXX-XXXX-XXXX"
      }
    }
  }
}

Restart Claude Desktop afterwards. The tools will appear prefixed with upservice_.

Optional: custom base URL

If Upservice ever changes the API host, or you use a private/on-prem instance, override it:

"env": {
  "UPSERVICE_API_KEY": "UPS-XXXX-XXXX-XXXX-XXXX",
  "UPSERVICE_API_BASE_URL": "https://public.upservice.io"
}

Testing locally

# uvx
UPSERVICE_API_KEY=UPS-XXXX uvx --from . upservice-mcp

# venv
UPSERVICE_API_KEY=UPS-XXXX venv/bin/python3 src/upservice_mcp/server.py

Or test interactively with the MCP Inspector:

npx @modelcontextprotocol/inspector uvx --from . upservice-mcp

Notes

  • Authentication uses a raw API key in the Authorization header (not a Bearer token) — this server handles that for you.
  • Destructive tools (delete_*, unassign_tag) are annotated with destructiveHint: true so MCP clients can warn users appropriately.
  • Some request bodies documented by Upservice have many optional fields; where a tool doesn't model a field explicitly, pass it via extra_fields (a plain JSON object) and it will be merged into the request body.
  • Never share your own UPSERVICE_API_KEY — each person should generate their own from their Upservice account settings.

推荐服务器

Baidu Map

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。

官方
精选
JavaScript
Playwright MCP Server

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。

官方
精选
TypeScript
Magic Component Platform (MCP)

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。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

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。

官方
精选
本地
TypeScript
VeyraX

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。

官方
精选
本地
graphlit-mcp-server

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。

官方
精选
TypeScript
Kagi MCP Server

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。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

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。

官方
精选