sbhooley/ainl-cortex

AINL Cortex is a Claude Code plugin that adds persistent memory and self-learning capabilities to the AI coding assistant. By default, Claude Code starts each session fresh with no memory of previous work. AINL Cortex solves this by storing every coding interaction, tool calls, decisions, errors, and outcomes, as typed nodes in a persistent knowledge graph. When a new session begins, the plugin retrieves the most relevant memories and injects them into context automatically, giving the assistant awareness of past work, your coding preferences, and previous mistakes to avoid repeating. The self-learning side works without making additional calls to a language model. Instead, it monitors patterns in your usage, successful workflows, recurring tasks, and failure modes, and promotes them into reusable knowledge over time. Goals set in one session persist across restarts and are automatically linked to new work as it progresses. The plugin also includes multi-agent coordination, allowing Claude Code to pass messages and tasks to other registered AI agents, and a mechanism for leaving yourself notes that surface at the start of the next session. A compaction safety feature ensures that when Claude Code summarizes a long session to free up context space, no buffered memory is silently lost before that summary is written. Installation is a two-step process, and the plugin is designed to degrade gracefully on errors, it never breaks the base Claude Code tool. It is built in Python and integrates with the AI Native Lang workflow system for additional token-saving optimizations. The full README is longer than what was provided.