# semantic-memory

ChromaDB-powered semantic memory server for nanobot.

Deployed as a **Custom MCP Integration** in the nanobot WebUI (Apps → Integrations → Custom). Once active, `@semantic-memory` in any topic gives nanobot semantic search across past conversations, health snapshots, and cross-silo topics.

## Quick Start

```bash
pip install -r requirements.txt
python -m prototype.server
```

Then add it in the WebUI as a stdio custom integration.

## What it stores

| Collection | Content |
|---|---|
| `conversation_memory` | Summarised conversation turns with dates and topics |
| `health_snapshots` | Daily Whoop + fasting + training metrics |
| `boss_brain_topics` | Discussion topics with statuses |

## Tools exposed via MCP

- `store(collection, text, metadata)` — embed and store text
- `search(collection, query, n)` — semantic search over stored content
- `store_conversation_summary(summary, date, tags)` — specialised for conversations
- `find_similar_days(metrics)` — retrieve health days matching current profile
- `collection_stats()` — count entries per collection