---
title: Conversation Memory System
emoji: 💾
colorFrom: blue
colorTo: green
sdk: gradio
pinned: false
---

# 💾 Conversation Memory System

SQLite-backed conversation tracker with text search, semantic search, and Markdown export. Fully offline.

## Features
- 💾 **SQLite persistence** — no external services
- 🔍 **Text search** — LIKE queries across all memories  
- 🧮 **Vector embeddings** — semantic search when `sentence-transformers` is installed
- 📤 **Export** — Markdown for any thread or all conversations
- 🔌 **Zero external APIs** — works fully offline

## Usage

### Save a message
Use the **Save** tab to store conversations from any source (user, assistant, system, agent-zero).

### Search
- **Text search**: Fast LIKE-based search across all message content
- **Semantic search**: Requires `sentence-transformers` (auto-detected)

### Export
Export individual threads or all conversations to Markdown.

## API Integration
```python
import requests

# Save a message
requests.post("https://scottzillasystems-conversation-memory.hf.space/api/save", json={
    "role": "user",
    "content": "Hello!",
    "thread_id": "my-thread"
})

# Search
requests.get("https://scottzillasystems-conversation-memory.hf.space/api/search?q=hello")
```