💾 Reflexive memory architecture — Daily logs are persistently stored, chunked, vectorized, and semantically retrieved for context-aware planning and regeneration.
🧠 Self-evolving logic — Recalls past decisions, codebases, and conversations to refine internal modules and regenerate stale logic over time.
📂 File-context-aware prompts — Injects scoped memory based on current file and intent — no redundant prompt entry.
🔍 Offline vector store — FAISS-based semantic index scoped by project, tag, and timestamp with pruning logic.
🧪 Under construction: Summarization engine, file watchers, and intent-based memory routing.
🛠 How to Use
Clone the repo. Launch Flask server. Load a code project. The assistant uses historical logs + semantic vector recall to assist coding sessions. Local-only for now.