Recursive Cognitive Optimization (RCO) (github.com)

Recursive Cognitive Optimization (RCO) has been introduced as a middleware orchestration tool that connects popular AI desktop applications like Claude Code and Codex, allowing them to function collaboratively under human direction. This innovation is significant for the AI/ML community as it streamlines the integration of existing AI tools without the need for additional API keys or pay-as-you-go costs, maximizing the utility of the subscriptions users already have. RCO's local design encourages a more cohesive workflow where models can propose, challenge, and refine ideas in coordinated cycles, enhancing the collaborative reasoning capabilities of AI systems. The RCO system features a visual Python coordinator that operates seamlessly within the Claude and Codex desktop environments. Users can initiate a collaborative process through an intuitive dashboard that manages connections and oversees the interaction between different AI agents. The middleware facilitates a structured workflow, offering modes such as human oversight, supervised reviews, and automated operations, while maintaining clear visibility of approval requests and task statuses. Although the current implementation focuses on a four-turn builder/reviewer workflow, there are plans for extensions, making RCO a versatile addition to desktop AI applications for more sophisticated, structured interactions.