A task list your agent ticks and you can read

Any MCP client and Pando, for a task list you both write in.

An agent working through a long job keeps a list somewhere. Usually that somewhere is inside the conversation, which means it is gone when the conversation is, and invisible while it runs.

Give a bullet the todo layout and its children are checkboxes, which the agent ticks as it finishes and you open in the same outline.

Set it up

Point the client at https://pando.ink/mcp, which speaks streamable HTTP.
Authenticate with OAuth 2.1 and dynamic client registration, or paste a key from Agents and API keys as a Bearer token.

How it is written

The layout is one property on one bullet, set in the same call that creates it, so a checklist costs no more than a plain list. Nothing moves in the graph: set the property back to bullet and the same content is an ordinary branch again.

create  text: "Migrate the billing tests"  layout: "todo"
        items: [ "Read the current fixtures",
                 "Write the new ones",
                 "Delete the old ones" ]

update  id: <the second one>  done: true

Why put it in your outline rather than in a file

You can tick one yourself, and the agent reads the tick next time it looks. A file the agent owns is a file you are only allowed to watch.
It sits next to the project it belongs to, not in a folder called tasks, so the list and the reasons behind it are one branch apart.
It survives the session. A new session reads the branch and carries on at the first unticked line.

One habit worth teaching it

Ask the agent to leave the lines it could not finish unticked, with the reason in the note. An agent that ticks everything and tells you afterwards which parts were guesses has given you a list you now have to audit. An agent that leaves three lines open has given you three lines to read.

Deeper

Connect an agent, about two minutes · The twelve tools · Which note apps an agent can reach · Who runs this