Cort

关于

Leveraging the Chain-of-Recursive-Thoughts (CoRT) methodology, this MCP server empowers AI to think more critically by simulating internal debates. Through iterative self-argument, the AI refines its responses, leading to more accurate, clear, and complete solutions. It enhances the original CoRT approach with features like multi-LLM inference and updated evaluation prompts.

主要功能

• Supports multi-LLM inference by randomly selecting different LLMs for each alternative.
• Implements Chain-of-Recursive-Thoughts (CoRT) methodology for enhanced AI reasoning.
• Provides different tool options to control output verbosity and evaluation prompt style.
• 1 GitHub stars
• Offers enhanced evaluation prompts to encourage AI to explain its reasoning.
• Configurable via MCP with logging options.

使用案例

• Improving the quality of AI responses by simulating internal debates.
• Exploring the diverse perspectives of different language models.
• Evaluating AI responses with enhanced prompts that encourage reasoning.