Library
A small space where we collect experiments and improvements made in software development by us and other talented companies in the space.
Scaling Long-Running Autonomous CodingJan 14, 2026
CursorCursor ran hundreds of AI agents in parallel for weeks, completing projects like a 1M+ line web browser and a Solid-to-React migration.
Why We Built Our Background AgentJan 12, 2026
RampRamp built Inspect, an internal coding agent that writes code and verifies its work through integrated tools. ~30% of their PRs are now written by it.