codec-cortex

What problem does it solve?

This Skill provides a universal communication protocol for LLM/SLM agents, enabling efficient transmission, storage, and reconstruction of agent state with minimal token overhead.

Core Features & Use Cases

• Semantic Compression: Compresses agent state (context, lessons, objectives, working memory) into a dense sigil format, significantly reducing the token overhead.
• Learning Engine: Automatically distills multiple specific lessons into general knowledge, providing a second-order compression that compounds across sessions.
• Layered Protocol: Consists of three independent layers: Knowledge, Transport, and Protocol.
• Use Case: Imagine an AI agent that needs to share its state with another AI. The CODEC-CORTEX Skill allows for efficient compression and transmission of this state, ensuring seamless communication between agents.

Quick Start

Initialize the CODEC-CORTEX Skill by running cortex learn init --workspace . and activate it by running cortex activate codec-cortex.