mirror of
https://github.com/samvallad33/vestige.git
synced 2026-04-25 00:36:22 +02:00
FSRS-6 spaced repetition, spreading activation, synaptic tagging, hippocampal indexing, and 130 years of memory research. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
1.8 KiB
1.8 KiB
Changelog
All notable changes to Vestige will be documented in this file.
The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.
[Unreleased]
Added
- FSRS-6 spaced repetition algorithm with 21 parameters
- Bjork & Bjork dual-strength memory model (storage + retrieval strength)
- Local semantic embeddings with fastembed v5 (BGE-base-en-v1.5, 768 dimensions)
- HNSW vector search with USearch (20x faster than FAISS)
- Hybrid search combining BM25 keyword + semantic + RRF fusion
- Two-stage retrieval with reranking (+15-20% precision)
- MCP server for Claude Desktop integration
- Tauri desktop application
- Codebase memory module for AI code understanding
- Neuroscience-inspired memory mechanisms:
- Synaptic Tagging and Capture (retroactive importance)
- Context-Dependent Memory (Tulving encoding specificity)
- Spreading Activation Networks
- Memory States (Active/Dormant/Silent/Unavailable)
- Multi-channel Importance Signals (Novelty/Arousal/Reward/Attention)
- Hippocampal Indexing (Teyler & Rudy 2007)
- Prospective memory (intentions and reminders)
- Sleep consolidation with 5-stage processing
- Memory compression for long-term storage
- Cross-project learning for universal patterns
Changed
- Upgraded embedding model from all-MiniLM-L6-v2 (384d) to BGE-base-en-v1.5 (768d)
- Upgraded fastembed from v4 to v5
Fixed
- SQL injection protection in FTS5 queries
- Infinite loop prevention in file watcher
- SIGSEGV crash in vector index (reserve before add)
- Memory safety with Mutex wrapper for embedding model
[0.1.0] - 2026-01-24
Added
- Initial release
- Core memory storage with SQLite + FTS5
- Basic FSRS scheduling
- MCP protocol support
- Desktop app skeleton