wuyoscar / Internal-Safety-Collapse

[NeurIPS 2026] Safety failures in coding agents and frontier LLMs. Reproducible workspaces for studying when AI agents generate unsafe data during task completion.

dream-num / univer

The Office Harness for AI Agents — Spreadsheets, Docs, Slides, Canvas, Relational Tables, and PDF in one runtime.

github / docs

The open-source repo for docs.github.com

excalidraw / cli

Official Excalidraw CLI for humans and agents