Cortex

About

Cortex is a sophisticated Model Context Protocol (MCP) server designed to facilitate advanced, multi-level reasoning. It exposes a powerful `reasoning.think` tool that enables AI agents to engage in structured thinking, maintain session continuity across complex tasks, and execute multi-step runs with progress notifications. With configurable depth levels (basic, normal, high) and flexible execution modes, Cortex empowers agents to tackle intricate problems, manage token budgets effectively, and access historical thoughts and session details through dedicated resource endpoints.

Key Features

• Multi-level reasoning (basic, normal, high) with configurable thought counts and token budgets
• Optional task execution with progress notifications for multi-step requests
• Dual execution modes: 'step' for single thoughts and 'run_to_completion' for batched thoughts
• Resource endpoints for viewing session lists, details, markdown traces, and individual thoughts
• Prompt and resource argument completion for seamless interaction
• 0 GitHub stars

Use Cases

• Building intelligent systems that require structured, continuous reasoning over time
• Automating sophisticated workflows where AI-driven thought processes need to be tracked and managed
• Enabling AI agents to perform complex, multi-step problem-solving