Features
A task manager your agents can actually use.
Six capabilities, each built for two audiences: the agent that does the work and the people who need to see it.
Persistent task lists
Tasks are stored in the hosted service the moment they're created. They don't live in the chat, so they don't disappear with it.
Your agent
- Reads the current list at the start of any new session
- Picks up work another session started
- Keeps one list going for weeks, not one conversation
Your team
- Sees the same tasks in the web view
- Opens the list without opening a chat
- Comes back tomorrow to the same list
Priorities and deadlines
Each task can carry a priority (urgent, high, normal, or low) and a due date, so the order of work is explicit rather than implied.
Your agent
- Sets priority and due date when it creates a task
- Asks the list for the most important open task
- Raises or lowers priority as plans change
Your team
- Sees what's due next
- Changes priorities directly
- Spots urgent items at a glance
Task updates
Agents create, edit, start, complete, block, and reopen tasks through clear operations. A blocked task records why it's blocked.
Your agent
- Lays out a multi-step plan in one go
- Marks a task in progress, done, or blocked
- Edits titles, notes, and fields without losing history
Your team
- Sees what changed and when
- Knows why something is stuck
- Reopens a task that wasn't really finished
Human visibility
A simple web interface shows open, in-progress, and completed work. It's the same data the agent uses, not a report generated after the fact.
Your agent
- Works on records people can inspect
- Leaves an activity history on each task
- Doesn't need to be asked for a status update
Your team
- Checks progress without interrupting the agent
- Adds, edits, and completes tasks directly
- Reviews finished work before signing off
Organized work
Group related tasks into lists, such as one per project, client, or recurring routine. Assignees and labels add another layer of structure.
Your agent
- Creates a new list for a new project
- Files tasks under the right list
- Filters by list, status, assignee, or label
Your team
- Keeps clients and projects separate
- Assigns tasks to people or to the agent
- Sees work grouped by list
Agent-friendly access
Compatible agents connect over the Model Context Protocol (MCP). The agent discovers the available task operations on connect, with no custom code.
Your agent
- Connects as a remote MCP server
- Gets task operations described in plain language
- Works the same way from any compatible client
Your team
- Sets up a connection in minutes
- Runs no server or database
- Switches agents without moving data
What Todos for Agents isn't
Knowing the edges helps you decide if it fits. For how the pieces connect, see how it works.
Not a project management suite
No Gantt charts, sprints, or resource planning. It's a focused task list, on purpose.
Not an agent
Todos for Agents doesn't do the work. Your agent does the work and keeps track of it here.
Not an automation engine
Tasks don't trigger workflows on their own. Agents and people decide what happens next.
Give your agents a place to get things done.
Stop losing track of work between conversations. Give your AI agents a persistent to-do list.