Vestige

About

Vestige is an advanced, 100% local memory server designed specifically for Claude, drawing on 130 years of cognitive memory research. It implements features like FSRS-6 spaced repetition, spreading activation, and synaptic tagging to create a memory system that intelligently retains and recalls information, preventing AI forgetfulness between sessions. By using prediction error gating and natural memory decay, Vestige ensures Claude's context remains relevant and efficient, allowing for a more human-like, adaptive interaction without token bloat or privacy concerns.

Key Features

• Persistent memory with intelligent retrieval for large language models.
• FSRS-6 decay mechanism that naturally fades unused memories, preventing context bloat.
• Prediction Error Gating automatically decides when to create, update, or supersede memories.
• Unified search tool offering keyword, semantic, and hybrid search capabilities.
• 100% local operation ensures data privacy and control for all memory interactions.
• 338 GitHub stars

Use Cases

• Providing Claude with relevant project or domain-specific context automatically.
• Enabling Claude to remember user preferences and instructions across multiple sessions.
• Setting reminders and future triggers for Claude to act upon proactively.