mcp-server-memo
A lightweight MCP server for managing session summaries and memos for LLMs like Claude.
Top Comments
MCP Server Memo is designed as a memory assistant for LLMs, allowing them to store and retrieve detailed session records through the MCP tool interface. The server preserves historical versions of sessions, organizes them chronologically, and uses local storage without requiring an external database. It is MCP compliant, optimized for performance, and has minimal dependencies, making it easy to maintain and extend.
Features
- Preserves History: All historical versions of a session are preserved.
- Time-Ordered: Sessions are organized chronologically.
- Local Storage: Uses the local filesystem for storage.
- MCP Compliant: Follows the Model Context Protocol specification.
- Performance Optimized: Optimized for file I/O and concurrent operations.
MCP Tools
- {'upsertSummary': 'Creates a new version of a session summary without deleting previous versions.'}
- {'getSummaryTool': 'Retrieves the latest version of a specific session summary.'}
- {'listSummariesTool': 'Lists available summaries with filtering, sorting, and pagination.'}
- {'updateMetadata': 'Updates metadata of the latest version of a session without changing the summary content.'}
- {'appendSummary': 'Appends content to a session summary, creating a new version.'}
- {'listAllSummariesTool': 'Lists all available summaries with basic information.'}
- {'getSessionHistory': 'Retrieves all historical versions of a specific session.'}
Usage with Different Platforms
client_workflow_example
javascript
import { v4 as uuidv4 } from 'uuid';
import { Client } from '@modelcontextprotocol/sdk/client/index.js';
import { StdioClientTransport } from '@modelcontextprotocol/sdk/client/stdio.js';
// Initialize client
const transport = new StdioClientTransport({
command: 'node',
args: ['dist/index.js'],
cwd: process.cwd()
});
const client = new Client({
name: 'example-client',
version: '1.0.0'
});
await client.connect(transport);
// Function to generate new session ID (done once per logical session)
function createNewSession() {
return uuidv4();
}
// Function to append to an existing session
async function appendToSession(sessionId, newContent) {
return client.callTool({
name: "appendSummary",
arguments: {
sessionId,
content: newContent,
title: "Example Session",
tags: ["example", "demo"]
}
});
}
// Function to get session history
async function getSessionHistory(sessionId) {
const response = await client.callTool({
name: "getSessionHistory",
arguments: { sessionId }
});
const result = JSON.parse(response.content[0].text);
if (result.success) {
return result.history;
}
throw new Error(result.error || "Failed to get session history");
}
// Usage example
const sessionId = createNewSession();
// Add initial content
await appendToSession(sessionId, "Initial conversation data");
// Add more content later in the conversation
await appendToSession(sessionId, "Second part of the conversation");
await appendToSession(sessionId, "Final part of the conversation");
// Get the full history
const history = await getSessionHistory(sessionId);
console.log(`Session ${sessionId} has ${history.length} versions`);
Related MCP Servers
View all knowledge_and_memory servers →Knowledge Graph Memory Server
by modelcontextprotocol
The Knowledge Graph Memory Server provides a simple solution for storing and managing persistent memory using a local knowledge graph, which is particularly useful for applications like chat personalization. It supports various operations for handling entities, relations, and observations, and offers flexible setup options, including Docker and NPX.
git-mcp
by idosal
GitMCP is a remote Model Context Protocol server that turns GitHub projects into documentation hubs for AI tools. It ensures accurate, up-to-date documentation access, minimizing code errors from hallucinations. The platform is open-source, free, and easy to integrate with various AI assistants, offering significant flexibility and utility for developers.
mindmap-mcp-server
by YuChenSSR
Mindmap MCP Server is a tool for transforming Markdown documents into interactive mindmaps. It supports multiple installation methods and works with various Model Context Protocol clients. The server is designed for both online and offline use, offering flexibility in how mindmaps are generated and viewed.