Run Claude Code and Codex sessions as spreadsheet cells for evals
Web page · https://agentusecases.ai/use-case/run-claude-code-and-codex-sessions-as-spreadsheet-cells-for-
Done with Claude Code, Codex. Author built a spreadsheet where each cell is a Claude Code or Codex terminal, used to run benchmarks such as a demo FDA submission pass/fail prediction bench.
First seen by Oriel on .
AI summary
recalc is an early-beta desktop spreadsheet that runs Claude Code and Codex sessions as manually triggered cells for agent evaluations and benchmarks.
- Cells contain Claude Code or Codex agents using @claude(...) or @codex(...), with references to other cells and an MCP server explaining the board.
- An FDA demonstration uses frozen submission dossiers, adding models as contestants and comparing their predicted decisions, but reports no accuracy or runtime results.
- Users can inspect an answer in a terminal, view API-priced run costs, and cap usage with Set Agent Budget.
- Agents use the operator’s Claude or ChatGPT plan; recalc is macOS-first, while Linux support is experimental and may contain bugs.
#claude code #codex #agent evaluations #spreadsheets
Generated by gpt-5.6-sol. Check the original for details.
Aliases
| Source | External ID | URL |
|---|---|---|
| Agent Use Cases | https://agentusecases.ai/use-case/run-claude-code-and-codex-sessions-as-spreadsheet-cells-for- | https://agentusecases.ai/use-case/run-claude-code-and-codex-sessions-as-spreadsheet-cells-for- Read discussion |