Thriving 
RS 
PY Persistent memory for AI coding agents — SQLite + FTS5 + vector search in a single file
https://archit15singh.github.io/posts/2026-02-28-designing-cli-tools-for-ai-agents/ ↗ HEALTH ? 78 / Stable
Thriving RS PY // readme
Memori
Persistent memory for AI coding agents — Rust + SQLite + FTS5 + vector search in a single file.
AI agents forget everything between sessions. Memori gives them a persistent memory layer that accumulates knowledge over time — so every session starts with the context it needs, not from zero.
pip install py-memori # 1. install (Linux/macOS/Windows, Python 3.9-3.13)
cd ~ && memori setup # 2. wire into Claude Code (writes ~/.claude/tools/memori/SNIPPET.md)
memori store "Chose SQLite over Postgres for zero-config portability." --meta '{"type":"decision"}'
memori search --text "database choice" # hybrid FTS5 + vector, auto-vectorized
Status: v0.7.1 released Sep 2026 (CHANGELOG) · Platforms: Linux, macOS (Intel + Apple Silicon), Windows · No API keys, no cloud, no LLM dependency
What you get
- Session continuity — your agent remembers the bug from yesterday, the architecture decision from last week, that you prefer Tailwind over Bootstrap.
- Zero config — one SQLite file stores text, 384-dim embeddings, JSON metadata, access stats.…