Gear is an open-source framework for recursive self-improvement of AI agent harnesses (prompts, tools, workflows, and context management). Built toward a scikit-learn for RSI, its modular architecture lets developers implement, compare, and extend optimization algorithms. With Gear, GPT 5.6 Luna reaches GPT 6 Astra-level performance: 88.9% vs. Astra + Codex’s 84.1% on AutomationBench’s public Marketing tasks.
Hitch is open-source infrastructure that makes agent harnesses interchangeable. Connect harnesses behind one interface and let an agent choose how to execute each task. Built for developers and researchers experimenting with multi-harness agent systems.
