What it does
recalc reimagines the spreadsheet for working with AI agents. Every cell can hold a number, a formula, a terminal command or an agent such as Claude Code or Codex, and agents write their plans and results into cells while you watch. Commands in other cells check their work and turn red when something is wrong, and you change things by editing a cell rather than interrupting a chat. Typical uses: put a cell that runs your tests next to the cell where an agent reports it is finished; put documents down the side and questions across the top so an agent answers every cell with a link to the exact passage; or give each agent its own row and collect everything that needs you in one Needs You list. A Driver agent can set up the board from a prompt, and results export to Excel. It runs locally with no account, server or tracking, and a board is just a folder of files. It works on macOS, with Linux experimental, and is free and open source under Apache 2.0. The authors call it early beta, so expect rough edges.
Who it’s for
Developers and analysts who run several AI agents and want to see their work in one place.
Use cases
- Running several coding agents side by side and checking their work with test cells
- Asking a hundred documents the same questions with cited answers
- Reviewing contracts or records on a grid and exporting to Excel


