Recalc is an experimental, open-source desktop spreadsheet designed as a shared workspace for people and AI coding agents. Cells can contain ordinary values, formulas, terminal commands, actions, or prompts for agents such as Claude Code and Codex. Formulas update when their inputs change; commands can rerun as inputs change; agent prompts run when requested. Agents can write plans and results into the grid, while users guide them by editing cells, reviewing outputs, or opening a cell’s agent session in a terminal. An MCP server lets agents working in their usual chat interact with the board.
The project’s central idea is to make agent work observable and checkable rather than bury it in separate chat sessions. Users can pair an agent’s completion claim with tests, arrange documents and questions in a grid to produce cited answers, coordinate several agents, or use formulas to validate financial data. A “Needs You” list gathers questions, errors, and completed work requiring attention. Boards save locally as folders, and results can be exported to Excel; agents can also use local models for sensitive documents. Recalc is free, but agents use the user’s existing provider accounts. It is in early beta, runs on macOS with experimental Linux support, and does not yet support Windows. Commands run with the user’s permissions, so the app is not a security sandbox; actions that change things require explicit user initiation.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.
Researchers tested 16 LLMs on all 55 TIS-100 puzzles, giving each one attempt and time to refine its code. The models found 133 new performance tradeoffs across 18 puzzles and set three outright cycle records, though human players improved on most solutions within days. (tis-100.wincon.dev)
Australia is considering a new criminal offence for AI companies that fail to take reasonable steps to prevent harm caused by their models. The government is also weighing an AI safety watchdog with power to impose significant fines, following a hack of a Medicare statistics portal involving OpenAI. (abc.net.au)
AI is increasingly becoming a gateway through which people find information. Audiences still resist having AI produce their news. (niemanlab.org)
Nvidia holds a central position in the AI industry, making disruption to the company potentially catastrophic for the wider ecosystem. The race for advanced AI is also intensifying economic and geopolitical conflict, while concentrated corporate power and infrastructure face growing political resistance. (france24.com)
LLM inference can require reading about 26 GB of model weights to generate each token from a 13-billion-parameter model. The explanation shows how model size drives memory bandwidth demands during text generation. (youtube.com)
Deletion Tuning matches full fine-tuning across three LLM post-training methods by zeroing fewer than 0.05% of model weights. Its compact masks also preserve prior-task accuracy better than full fine-tuning, LoRA, and random sparse tuning. (arxiv.org)
Today's best Hacker News stories, summarized and screenshotted, one email a day.