Teamwork MCP
Teamwork.com tasks, Desk tickets, Spaces pages and chats for an agent: a hosted server with role profiles and a read-only mode
- 28 ★ on the repository
- Last commit
- MIT
Broad access
Why this level
- The agent creates, edits and may delete tasks, tickets and pages in the workspace
- Chats and tickets bring correspondence and customer data into context
Install
- Works in
Manual install
claude mcp add --transport http teamwork https://mcp.ai.teamwork.com/Then run /mcp and sign in to Teamwork.
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- The agent creates, edits and may delete tasks, tickets and pages in the workspace
- Chats and tickets bring correspondence and customer data into context
This is third-party code. Review the repository files before installing.
What it does
The official MCP server of Teamwork.com, a project management service. The agent works with projects, tasks, people, timers, Desk support tickets, Spaces pages and chat conversations. The full list of create, read and update operations is in the repository's tool reference. The hosted server at https://mcp.ai.teamwork.com has role profiles so that not every tool floods the context: project-manager, support, analyst, knowledge-manager and ops with the full set. Sign-in works through browser OAuth2 or a token. You can also run the server yourself, written in Go: in HTTP mode for shared deployments or STDIO mode for the desktop. A read-only option limits tools to reading data, which is safer for a first try.
Who it is for. Project managers and support teams on Teamwork.com who want to run tasks and tickets through conversation with an agent.
Good fit when
- You need a summary of a project's overdue tasks with assignees
- Support tickets are triaged in a batch: the agent groups them and suggests replies
- Time entries must be rolled into a weekly report
- Knowledge base pages in Spaces get tidied together with the agent
- The role is narrow and a profile with few tools fits
Not a fit when
- Your team works outside Teamwork.com: the server does nothing else
- An administrator has not enabled MCP in site settings: the connection will not pass without it
- You need changes with no oversight: some operations delete and edit data, and read-only mode is for checking
- Company policy forbids giving a token to an AI client
Example request
Show the overdue tasks of the Website Launch project with assignees and due dates.MCP
- Transport
- http, stdio
- Authentication
- OAuth
Environment variables
TW_MCP_BEARER_TOKENsecret- Token for STDIO mode
Limitations
A site administrator must enable MCP under Settings, AI, and the user needs permission to create an API key. Running it yourself needs the Go version named in the README.
Profiles shorten the tool set, but different profiles see different data, so the analyst report does not replace project-manager work. List output schemas mark every field optional, so OpenAI's strict structured output mode does not fit them.
Availability of Teamwork.com from Russia was not checked.
How to disable. Run claude mcp remove teamwork and revoke the API token in the Teamwork account. An administrator can switch MCP off for the whole site in settings.
FAQ
How do I connect Teamwork to Claude Code?
With claude mcp add --transport http teamwork https://mcp.ai.teamwork.com/ and sign-in through /mcp. Instead of OAuth you can pass a token in the Authorization header.
What are profiles?
Addresses that open only part of the tools for a role: project-manager, support, analyst, knowledge-manager and ops with the full set.
Which Teamwork products are covered?
Projects, Desk, Spaces and Chat.
Can the agent be limited to reading?
Yes, the server has a read-only mode. It suits a first check on live data.
Is anything needed on the Teamwork side?
Yes, an administrator enables MCP in Settings, AI. The user also creates an API key.
Discussion
No questions or reviews yet.
Related
MCP serversThe official Taskade MCP and plugin: an agent manages workspace projects, tasks, automations and AI agents.
The official Todoist MCP server: the agent finds, creates and updates tasks and projects in your account
Hosted Workopia MCP server: job search in more than 90 countries, resume tailoring for a posting, cover letters and an application dashboard
The Mindwtr task manager MCP server: an agent reads GTD tasks, projects and areas, and writing needs a separate flag.
Jira MCP in Go: an agent searches issues by JQL, runs sprints, comments and worklogs and sees an issue's branches and pull requests on Cloud and Server
Plane MCP Server: project management in Plane
Plane MCP Server
Official Plane MCP server: the agent manages projects, work items, cycles and releases through 30 tools
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| Teamwork MCP (this card) | yes | OAuth | 28 | |
| monday.com MCP | yes | OAuth | 437 | monday.com boards and items, while Teamwork adds Desk and Spaces |
| Atlassian Rovo MCP Server | yes | OAuth | 1,160 | Jira and Confluence in the Atlassian ecosystem |
| Linear MCP Server | no | API key | 346 | the Linear issue tracker for engineering teams |
README in short
The docs describe three ways to run it (HTTP, STDIO and a test CLI), the repository layout and development commands. Step-by-step client setup lives in the usage guide with separate pages for ChatGPT, Claude, Gemini CLI and VS Code.