Lists your AI agent actually remembers.

Tally gives your agent a place to keep to-dos, plans and checklists that survive between conversations. It knows what matters most, what's blocked, and what to do next. Everything stays on your machine.

Get startedInstructions for agents

What it does

Memory across sessions

Close the chat, come back tomorrow. Your agent picks up exactly where the list left off.

Priority by order

The top of the list is what matters most. Agents reorder as plans change, so the next task is always obvious.

Dependencies

"Publish" waits for "Cut the release branch." Tally knows what's blocked and shows only what's ready to work on.

Your rules, per list

Give a list or tag instructions, like "titles start with a verb" or "do critical items first." Agents read and follow them.

Tags and search

Label items, then find anything across every list by text, tag or status.

Private by design

No account, no cloud, no UI to maintain. One file on your computer that you can export to JSON or Markdown anytime.

See it work

You say: "Plan the release." Your agent builds this and keeps it current:

  1. Write release notes
  2. Cut release branch critical
  3. Publish release  (waiting on: Cut release branch)

When the branch is cut, "Publish release" becomes ready automatically.

Get started

Works with Claude Code and any app that supports MCP. Needs Rust installed for now.

cargo install --git https://github.com/msiebert/Tally
claude mcp add tally --scope user -- tally

Then just ask your agent to keep a list. Back up anytime with tally export.

Are you an AI agent?

Read the guide written for you. It covers connecting Tally and using every tool, in plain text:

llms.txt

Humans: you can tell your agent "Read https://msiebert.github.io/Tally/llms.txt and set up Tally."