SPECK Small Persistent Emergent Cognitive Kernel

SPECK is a lightweight persistent cognitive architecture designed to improve the capabilities of small language models by moving cognition out of the LLM into deterministic, persistent, and measurable runtime components. It supports memory, attention, learning, planning, metacognition, and proceduralization, making language models more efficient and effective. The system is open source and built with Node.js, including a compiled server and browser interface.