shefyyuri/xai-cortex

xai-cortex is a Python library for managing context across conversations with large language models (LLMs, AI systems like GPT or Claude). The core problem it addresses is that LLMs have a limited "context window", they can only consider a certain number of words (measured in tokens) at a time. In a long or complex conversation, important earlier details may get cut off because there is not enough room. This library lets you organize conversation context as named segments, each with a priority level. When the total content would exceed the token budget you configure, the library automatically trims or compresses lower-priority segments first, preserving the most important information. Compression works by summarizing low-priority segments to make them smaller rather than discarding them entirely. You can also save the full context state as a snapshot and restore it later, enabling memory that persists across separate sessions. A timeline feature allows rolling back to a previous context state. You would use this if you are building an AI-powered application where the conversation or agent needs to remember things over a long period, for example, a chatbot that recalls user preferences, an AI game character that maintains personality and memory, or an agent that needs to track a complex ongoing task. The library handles both in-memory storage (for a single session) and SQLite (a simple file-based database, for persistence between sessions).