Memories Lane

A flexible MCP server designed to facilitate sharing conversation context across different interactions with Large Language Models (LLMs).

Features

• Easy-to-use tools for persisting and retrieving conversational context.
• Customizable prompts for efficient workflow integration.
• Support for structured content including Mermaid diagrams, code snippets, and tasks.

Tools

Usage:

Tools are automatcally called by the LLM prompts. You do not need to call them directly.

leave_off

Usage:

• Automatically called via MCP prompts with the required context parameter.

Stores the current conversation context and returns a unique identifier (id) for later retrieval.

pick_up

Retrieves previously stored context using the unique identifier obtained from leave_off.

Usage:

• Automatically called via MCP prompts with the required id parameter.

list_context_items

Retrieve a list of all stored context items ordered by their creation

Usage:

• Automatically called via MCP prompts.

remove_context_item

Delete a specific context item using its unique identifier.

Usage:

• Automatically called via MCP prompts with the required id parameter.

Prompts

Customize your conversational workflows easily using prompts.

In Claude Desktop, activate by clicking "Attach to MCP", then select your desired prompt from the "Choose an integration" menu.

💾 Save

Summarizes the current conversation and utilizes the leave_off tool to store the context.

Optional Parameters:

• projectType: Define the project type. Available options:
  • general (default)
  • technical
  • creative
  • problem-solving
• includeMermaid: Include Mermaid diagrams (yes or no, default: no)
• includeCode: Include code snippets (yes or no, default: no)
• includeTasks: Include task lists (yes or no, default: no)

📂 Continue

Loads previously saved conversation context using the pick_up tool.

Required Parameter:

• id: Unique identifier generated by the leave_off tool.

📂 Resume

Loads most recently saved convesation context.

Installation

Prerequisites

• Node.js
• MCP-compatible client (e.g., Claude Desktop)

Setup

Integrate into your MCP client configuration:

Using NPX:

{
  "mcpServers": {
    "memories-lane": {
      "command": "npx",
      "args": ["-y", "memory-lane", "<path-to-context-storage>"]
    }
  }
}

Local Development:

{
  "mcpServers": {
    "memories-lane": {
      "command": "<path-to-node>",
      "args": ["<this-directory>/stdio.mjs", "<path-to-context-storage>"]
    }
  }
}

Replace <path-to-context-storage> with your desired storage directory path.

License

MIT