Show HN: Raven – The harness of harnesses, built for RSI
Summary
The article presents Raven as an open source, multi-agent harness designed for recursive self-improvement. It describes built-in agents (Raven-Research, Raven-Code, Raven-Design, Raven-Oncall), integration with third-party agents, memory and planning components via EverOS, and a web UI with showcases including game development, RSI experiments, and a product launch. It highlights Raven as a framework for autonomous AI tasks and multi-agent orchestration.