← Back to directory
T

Todoist MCP Server

Official
Official MCP server to connect AI agents to Todoist.
Category
Collaboration #47 of 55
Stars
★ 554 Very popular
Transport
Streamable HTTP · stdio (local process)
Runtime
Node.js · Hosted (no local runtime)
Credentials
API key / credential required
License
MIT
Last commit
Tools
5
40FMRS · D

The Todoist MCP server is officially maintained and offers reliable task management integration. Its design emphasizes reusability and complete workflows rather than isolated atomic actions. It supports multiple popular clients and is easy to install and configure. Suitable for individuals and developers, but attention should be paid to permissions and data security.

Strongest · Setup experience 12/20 Weakest · Reliability 5/20

Reliability
5/20
Security and permissions
6/20
Maintenance
7/20
Documentation
10/20
Setup experience
12/20
Why each score
Reliability 5/20
Only the README and repository metadata were supplied; no source code, CI logs, or tests were available, so clean startup, the MCP init/tool-list handshake, and tool/behavior consistency cannot be verified. The README claims npm start, npx @doist/todoist-mcp, and client integrations, but those are untrusted claims under the evidence rules. No contradiction is visible, but key paths have no supporting evidence, hence 5.
Security and permissions 6/20
The README describes conventional credential handling via a TODOIST_API_KEY environment variable and browser-based OAuth; no real secrets, malware, or covert exfiltration indicators appear, so no red-line score is triggered. However, source-level review was impossible and least privilege, data boundaries, and confirmation for write tools such as addTasks are not established. The README only advises checking the account before writes and using a test account, so 6.
Maintenance 7/20
The repo is not archived, has an MIT license, and the README describes release-please and a publish workflow, which are positive maintenance signals. But no commit history, release history, issue-response metrics, or dependency-update evidence was supplied. Nine open issues alone do not prove responsiveness; under missing-evidence rules the score is 7.
Documentation 10/20
The README covers installation, authentication, and concrete client configuration for Claude Desktop, Cursor, Claude Code, and VS Code, and references layered docs under docs/. However, complete tool parameters, limits, costs, and troubleshooting are absent from the supplied material, and the referenced doc contents were not part of this review, so 10.
Setup experience 12/20
The README provides a short path from npx @doist/todoist-mcp or the remote HTTP URL to configuration snippets for four mainstream clients. The steps appear clear and few, but the static calibration caps setup at 15 without verifiable execution evidence from files. Since no CI or tests were supplied to confirm the handshake, the score is 12.

Static review · not runListed 2026-08-07

Read the FMRS scoring method →

Fit and risk

What it can accessUses the networkChanges third-party account data

Best for

  • Users who need AI assistants to directly operate Todoist tasks.
  • Developers who want seamless task management in AI tools like Claude, Cursor, VS Code.
  • Developers who want to build custom AI workflows with reusable tools.

Not for

  • Users who do not need AI task management features.
  • Enterprises requiring multi-user collaboration or permission management (this server operates on a single user account).
  • Users with strict data privacy requirements (third-party service may need evaluation).

Required permissions

  • Requires a valid Todoist API key (TODOIST_API_KEY) or OAuth authentication.
  • Tools can read and modify user's Todoist account data, including tasks, projects, and labels.

Risks and side effects

  • AI agents may accidentally delete or modify tasks; it is recommended to use a test account or temporary project.
  • Leaked API keys or OAuth tokens could lead to unauthorized account access.
  • As it relies on external services, there is potential for downtime or latency.

Setup

Before you start

Runtime:Node.js · Hosted (no local runtime)

TODOIST_API_KEY requiredsecret Todoist API key generated from your Todoist developer/integration settings; required for running tools directly (npm run tool) or as a library, while the hosted MCP endpoint uses OAuth instead.
  1. Quick start with npx: npx @doist/todoist-mcp
  2. For Claude Desktop: Settings → Connectors → Add custom connector, enter https://ai.todoist.net/mcp and complete OAuth authentication.
  3. For Cursor: create configuration in ~/.cursor/mcp.json or project-specific .cursor/mcp.json (see install_config).
  4. For Claude Code: use the official plugin with /plugin marketplace add doist/todoist-mcp and /plugin install todoist@doist, or manually run claude mcp add --transport http todoist https://ai.todoist.net/mcp.
  5. For VS Code: Command Palette → MCP: Add Server, select HTTP transport, use the provided JSON configuration.
  6. For other clients: run npx -y mcp-remote https://ai.todoist.net/mcp.
claude_desktop_config.json
{
  "mcpServers": {
    "todoist": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://ai.todoist.net/mcp"
      ]
    }
  }
}

Shown for Claude Desktop. Other clients may use a different file or key (VS Code uses "servers") — the configurator below converts it.

.vscode/mcp.json
{
  "servers": {
    "todoist": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://ai.todoist.net/mcp"
      ]
    }
  }
}

Goes in your project's .vscode/mcp.json (VS Code uses a "servers" key).

Terminal
claude mcp add todoist -- npx -y mcp-remote https://ai.todoist.net/mcp

Run it in a terminal; replace any <…> placeholders with your own values first.

Check that it works

Check that the todoist server's tools (e.g. findTasksByDate, addTasks, search, fetch) appear in your client's tool list; the first Todoist tool call triggers OAuth in your browser, and successful authentication confirms the connection.

Troubleshooting

  1. Ensure TODOIST_API_KEY is set correctly; use `npm run tool -- user-info '{}'` to verify connected account.
  2. Check network connectivity and firewall to ensure access to `https://ai.todoist.net/mcp`.
  3. For write operations, use a test environment first to avoid accidental data modification.

Things to try

Once connected, you can ask your AI assistant things like:

  • Find my tasks by date
  • Add a task for today: buy milk
  • Search for tasks containing 'report'
  • Fetch details of this task

Tools 5

search read-only
Follows the OpenAI MCP specification, performs search operations, returns JSON-encoded results.
fetch read-only
Follows the OpenAI MCP specification, fetches content, returns JSON-encoded results.
findTasksByDate read-only
Finds tasks by date.
addTasks writes
Adds new tasks.
other_tools_in_src_tools
More tools are available in the src/tools directory; see the full list.

Use cases

Let AI assistants help users manage tasks in Todoist, such as searching, creating, and modifying tasks.
Interact with Todoist account via natural language with AI.
Integrate Todoist tools into custom AI applications for task automation.

Supported clients

Claude Desktop
Cursor
Claude Code
VS Code

Listed from the project's documentation, not tested by this site.

Overview

An official MCP server by Doist that allows AI agents to access and modify a Todoist account on behalf of the user. Provides a set of reusable tools that can be used via MCP or integrated directly into custom AI conversational interfaces.

Similar servers

Source revision c9bcda17a6d8 Data synced 2026-10-11 Read the FMRS scoring method