Update README.md

This commit is contained in:
2026-06-02 07:05:45 +00:00
parent 7d1bdd48c2
commit 0e3ba467f2
+48 -2
View File
@@ -1,3 +1,49 @@
# task-69f8e929da860fb4533faa2a # MCP Memory Server
MCP-сервер для управления памятью агента This repository contains a simple **MCP (Model Context Protocol)** server that provides a persistent key/value store with optional namespace support. The server is built with **FastMCP** and exposes tools that can be called by other agents or CLI utilities.
## Features
- **Basic CRUD** operations: `save`, `get`, `delete`, `list_keys`.
- **Namespace support**: `save_with_namespace`, `get_by_namespace`, etc.
- Data is persisted to `memory_data.json` with timestamps.
- Simple client demo (`memory_client.py`).
## Installation
```bash
pip install -r requirements.txt
```
## Running the Server
```bash
python memory_server.py
```
The server will start listening via **stdio** transport.
## Running the Client Demo
```bash
python memory_client.py
```
You should see output similar to:
```
Сохранено: True
Данные namespace 'default':
default:username: Алексей
Ключи с 'name': ['default:username']
```
## How It Works
- The server uses a JSON file (`memory_data.json`) to store data. Each entry is a mapping of `namespace:key` to a dictionary containing `value` and `timestamp`.
- All tools are registered with `@mcp.tool()` and can be invoked by any MCP client.
- The client uses `fastmcp.Client` to connect via stdio and call the tools.
## Extending
You can add more tools or modify the storage backend (e.g., switch to a database) by extending the `MemoryServer` class.