TL;DR:
Connecting Claude and ChatGPT to Todoist via MCP lets your AI manage your task list directly from chat.
Manual setup is possible using local config files and raw API tokens, but it requires you to manage credentials and handle token expiry yourself.
Composio provides a secure, managed alternative with a guided connection flow and automatic token refresh.
Composio handles the integration infrastructure for you, including a Tool Router that directs each request to the right connected app.
Composio's free tier includes 20,000 tool calls per month with no credit card required.
Most AI assistants can tell you what to do. With MCP, they can also do it. Connecting Todoist to Claude Code, Cowork or ChatGPT lets your AI create tasks, update projects, organize priorities, and turn emails, meetings, and documents into actionable work. This guide shows how to set it up with Composio and use it to delegate real workflows end to end, not flag what to do.
If you're comfortable with the terminal, use Claude Code. Otherwise, Claude Cowork or ChatGPT Work requires no command line at all.
How to connect Todoist to ChatGPT Work
Prerequisites
A ChatGPT account
Access to the Todoist workspace you want to connect
A Composio account
A Todoist account ready to authenticate via OAuth
How to open the MCP settings
Open ChatGPT Work → Settings → Plugins → MCP.
How to add the Composio MCP server
Click Add server → enter Composio as the server name → select Streamable HTTP as the type → paste https://connect.composio.dev/mcp into the URL field → click Save.
How to authenticate your Todoist account
Click Authenticate → sign in with your Composio account → approve the requested OAuth permissions by selecting Allow access.
How to start using Todoist with ChatGPT Work
Ask ChatGPT Work to perform Todoist actions using natural language → authorize any requested actions → manage tasks, projects, and comments directly from ChatGPT Work.
How to connect Todoist to Claude Code
Prerequisites
A Claude Code installation
A Composio account
A Todoist account
Add the Composio MCP to Claude
Click Generate MCP URL on the Composio page → copy the generated MCP configuration command → run it in your terminal to add Composio to Claude Code.
Start Claude Code
Open your terminal → run the following command:
claudeOpen your MCP list
In Claude Code → enter the following command:
/mcpSelect Composio and authenticate
Find Composio in the MCP list → select it → click Authenticate.
Authorize Composio
Claude Code redirects you to the Composio OAuth page → authorize Composio → complete the authentication flow → return to Claude Code. Your Todoist integration is now ready to use.
How to connect Todoist to Claude Cowork
Prerequisites
Claude Desktop with access to Cowork
A Composio account
A Todoist account
Open Customize
Open Claude Desktop → click Customize in the left sidebar → select Connectors → click the + icon at the top.
Add the Composio MCP server
Click Add custom connector → paste the following Composio MCP server URL:
https://connect.composio.dev/mcpAuthorize Composio in your browser
Click Connect → Claude redirects you to a browser window → authorize Composio to continue.
Connect your Todoist account
Return to Cowork → ask the agent to connect to Todoist or give it a Todoist-related task → follow the prompt to authenticate and authorize access to your Todoist account.
For example, ask Cowork to add a high-priority task for today, create a new project called Team Offsite, or close all completed tasks from this week.
Practical ways to delegate tasks to your AI
Turn your inbox into a self-managing work queue
Every email in your inbox is a decision you haven't made yet. Reply now? Delegate it? Follow up next week? That stack of decisions is what makes email exhausting, not the volume.
An AI connected to Todoist can make those decisions for you. Instead of simply creating tasks from emails, it can distinguish between newsletters, FYIs, approvals, customer requests, invoices, and conversations that require follow-up. Tasks get created only when work needs to happen, get assigned priorities based on urgency, and arrive with the original email attached for context. Follow-up reminders appear automatically if someone hasn't replied within a set timeframe.
By the time you open Todoist, your inbox has already been converted into an organized execution plan instead of an overwhelming list of unread messages.
Example prompt
Review every email I've received today. Ignore newsletters and FYIs. Create Todoist tasks only for emails that require action, estimate their priority based on the sender and deadline, attach the original email, and remind me to follow up after five days if nobody replies.
Turn every meeting into an executable project before anyone leaves the room
In practice, meeting momentum fades fast, usually within minutes of the call ending. By the time everyone returns to their inbox, half the discussion is already gone.
An AI connected to your meeting transcripts, calendar, and Todoist can eliminate that gap entirely. As soon as the meeting finishes, it can extract every decision, identify every commitment that was made, separate discussion from actual action items, create tasks, group them into project sections, assign realistic due dates based on dependencies, and prepare the project before momentum fades.
Instead of spending another half hour writing meeting notes, the team can immediately start working.
Example prompt
Read today's product planning meeting transcript. Ignore discussion and capture only confirmed action items. Create a Todoist project called "Q4 Product Launch," organize tasks into Design, Engineering, Marketing, and QA sections, estimate realistic due dates based on dependencies, and flag any decisions that still need clarification.
Break massive documents into work your team can actually execute
A product requirements document might be fifty pages long. A client proposal could span dozens of deliverables. Reading them is easy. Converting them into an execution plan is what takes hours.
Your AI can identify every deliverable, milestone, dependency, and stakeholder, then automatically generate a structured Todoist project. Instead of ending up with one vague task like "Launch website," you'll have dozens of clearly defined pieces of work arranged in the order they actually need to happen.
Implementation guides, onboarding documents, technical specifications, marketing campaigns, and operating procedures all have the same problem: they describe what needs to happen without breaking it into work someone can pick up and start.
Example prompt
Read this product requirements document and convert it into a complete Todoist project. Break every deliverable into actionable tasks, identify dependencies, organize work into logical phases, estimate realistic deadlines, and include the relevant section of the document inside each task description.
Plan your day around reality instead of wishful thinking
A task list built in the morning reflects your intentions, not your calendar. Then meetings appear, priorities shift, and half the list rolls over to tomorrow untouched.
Your AI can look at your calendar, your existing Todoist projects, upcoming deadlines, and even your working hours before deciding what should actually get done today. It can automatically postpone less important work, reserve uninterrupted focus blocks, and schedule shallow work between meetings, so that high-impact tasks get done while you still have energy.
Instead of asking "What should I work on next?" you always have a realistic plan that adapts as your day changes.
Example prompt
Review my calendar, Todoist, and upcoming deadlines. Build the most realistic schedule for today, reserve uninterrupted focus time for deep work, move anything that won't realistically fit into tomorrow, and explain why you prioritized each task.
Build recurring operating systems instead of recurring reminders
Recurring tasks usually remind you that work exists. They rarely tell you everything that needs to happen.
Your AI can create entire recurring workflows. A weekly client review can automatically generate agenda preparation tasks three days beforehand, remind you to collect analytics the day before, create follow-up tasks after the meeting, and schedule check-ins for unresolved action items.
Instead of remembering processes, you simply execute the work that appears.
Example prompt
Create a recurring workflow for my weekly leadership meeting. Three days beforehand, remind me to collect KPI reports. One day beforehand, create an agenda review task. After every meeting, remind me to capture action items and schedule follow-ups for unresolved decisions.
Turn conversations into projects without taking notes
Ideas don't only happen during meetings. They happen in Slack, Teams, Discord, customer calls, WhatsApp messages, and quick conversations throughout the day.
Instead of manually writing everything down later, your AI can monitor those conversations, identify commitments as they're made, separate ideas from actual decisions, and continuously update your Todoist projects without you lifting a finger.
That means action items never disappear inside long message threads, and projects stay current even when work is happening across multiple communication tools.
Example prompt
Review today's Slack conversations across the Product and Marketing channels. Create Todoist tasks only for confirmed action items, include links back to the original conversations, group related work together, and flag anything that appears blocked waiting for someone else.
Continuously reorganize your priorities as work changes
No project stays perfectly organized for long. Deadlines move, blockers appear, priorities shift, and tasks that mattered yesterday become irrelevant today.
Rather than maintaining Todoist manually, your AI can continuously audit your workspace. Overdue work gets flagged, abandoned tasks archived, duplicates merged, and priority changes recommended based on upcoming deadlines. Bottlenecks surface before they block progress, and projects consuming attention without output get called out.
Instead of managing your task manager, your task manager manages itself.
Example prompt
Review every Todoist project. Archive stale tasks older than 90 days, merge duplicates, identify overdue work, suggest priority changes based on upcoming deadlines, and recommend the five highest-impact tasks I should focus on this week.
Sign up with Composio and start connecting Todoist to your AI assistant. It's free, and no credit card is required.
FAQs
How much does it cost to run the Todoist MCP setup?
The manual setup is free but requires you to manage credentials and keep configuration files up to date yourself. Composio's managed setup is free for up to 20,000 tool calls per month with no credit card required, and paid plans start at $29/month for 50,000 tool calls.
Can I manage the permissions and scopes for Todoist while using Tool Router?
You can configure which Todoist scopes and actions are allowed when connecting your account to Composio. You can also bring your own OAuth credentials or API configuration so you keep full control over what the agent can do.
Is it safe to store my Todoist API token in a local JSON file?
Storing tokens in local config files carries credential risk, especially on shared or cloud-synced machines. Composio's managed auth layer removes this risk. Composio is SOC 2 and ISO 27001 certified, so you're not the one holding the raw token.
Key terms glossary
Model Context Protocol (MCP): MCP is a universal connector for AI: it lets a model like Claude or ChatGPT read and write data in your everyday apps through one consistent format, instead of a custom integration per app.
Tool Router: A Composio feature that directs AI requests to the correct application based on the user's active connections. When an agent needs to create a task, the router determines whether to use Todoist, Asana, or Linear based on what you've connected, without any conditional logic in your setup.
API token: A secure key that external services use to verify your identity when making programmatic requests. For Todoist, your API token grants full read and write access to your task data.
In-chat authentication: A Composio feature that lets AI agents prompt you to connect new accounts mid-conversation using a Connect Link URL. You authorize once and the agent continues without interruption.
LLM (Large Language Model): An AI model trained on large amounts of text data that can understand and generate human language. Examples include Claude, GPT-5, and other conversational AI systems that power chat assistants.