Your AI assistant, working from your real task list
Taskeni runs a remote MCP server, so assistants like Claude, ChatGPT, Cursor, and Claude Code can read your boards and update tasks as you. You approve each connection, choose read or read & write, and can disconnect it at any time.
Works with your assistant
Claude, ChatGPT, Cursor, Claude Code, and other MCP clients.
OAuth or API key
Approve in the browser, or create a key for tools that can't sign in.
Ten narrow tools
Five reads and five writes. Nothing can delete.
Acts as you
An assistant sees only the boards you can see.
What MCP is, in one paragraph
The Model Context Protocol (MCP) is an open standard that lets an AI assistant use tools from other apps. Once you connect Taskeni, you can ask your assistant “what's due for me this week?” or “file these three bugs on the Sprint board” and it calls Taskeni directly instead of guessing or asking you to copy and paste.
Connect in a minute
Copy the server URL from Settings → API & AI agents. When an assistant connects, Taskeni opens a consent page at app.taskeni.app/connect where you approve it and choose read only or read & write access.
- **Claude:** Settings → Connectors → Add custom connector, paste the URL, then approve it in Taskeni.
- **ChatGPT:** turn on developer mode in ChatGPT's settings, create a connector with the URL, and approve it in Taskeni.
- **Claude Code:** run `claude mcp add --transport http taskeni <server URL>`, then `/mcp` to sign in through your browser.
- **Cursor:** add the server URL to `.cursor/mcp.json`.
- **Anything else:** create an agent API key in the same settings tab and send it as a Bearer token. The key is shown once.
The ten tools
Five tools read and five write. There is deliberately no delete, archive, bulk, member, or invitation tool.
- **list_my_tasks:** your open tasks, soonest due first, filtered by today, this week, overdue, or board.
- **search_tasks:** full-text search across task titles, descriptions, and comments.
- **get_task:** one task in full: description, checklists, subtasks, blockers, file names, linked pull requests, and recent comments.
- **list_boards:** your workspaces, and the boards and lists you can see.
- **get_board_summary:** counts per list, open and done, overdue, due this week, and recent activity.
- **create_task:** a new task on a board, with assignees, due date, priority, and labels.
- **update_task:** change a task's title, description, assignees, dates, priority, or labels; add checklist items or tick them off.
- **move_task:** move a task to another list, or another board in the same workspace.
- **add_comment:** post a comment as you, with @mentions.
- **complete_task:** mark a task done, or reopen it.
Built to be safe to connect
An assistant acts as the person who connected it and goes through the same permission checks as the web app, so it can't see private boards you're not on. Nothing it can do deletes data. Tasks it creates say “via AI assistant” in their activity. Requests are rate-limited per connection, and the server tells assistants to treat task titles, descriptions, and comments as data, never as instructions.
- See every connected assistant and API key under **Connected agents**, with when it was added and last used.
- Disconnect any of them in one click; access ends right away.
Common questions
Which assistants work with Taskeni?
Any client that supports remote MCP servers over HTTP, including Claude, ChatGPT (with developer mode), Cursor, and Claude Code. Settings → API & AI agents has setup steps for each.
Is the MCP server on the Free plan?
Yes. The MCP server is on every plan, and it doesn't use Taskeni's AI credits: the assistant runs on your own Claude, ChatGPT, or Cursor account. See pricing.
Can an assistant delete my tasks?
No. There's no delete or archive tool. The worst a write-enabled assistant can do is change fields, move tasks, or post comments, all of which show in the task's activity.
What's the difference between OAuth and an API key?
With OAuth, the assistant opens Taskeni in your browser and you approve it there. An agent API key is for tools that can't open a browser; you create it in Settings and paste it into the tool. Both act as you, and both can be disconnected.
Can I give an assistant read-only access?
Yes. Choose Read only when you approve the connection or create the key. Read-only connections can use the five read tools only.
Where can I learn more?
See the help center for setup guides, and our post on connecting AI assistants with MCP for example prompts.