Cursor10x Memory System

Cursor10x Memory System

A comprehensive memory system for Cursor using the Model Context Protocol (MCP).

🚀 ANNOUNCING CURSOR10X SYSTEM 🚀

Transform Your Development Process with AI-Powered Autonomous Systems

🔥 The Cursor10x Memory System is now part of the complete Cursor10x Platform! 🔥

Discover the full autonomous development ecosystem at GitHub featuring:

📋 Task Management System - Guided implementation with step-by-step tasks 🔄 Autonomous Memory - Context-aware AI that remembers your entire project 📊 Project Blueprints - Complete technical architectures created for your specifications 📁 File/Folder Architecture - Optimized project structure with best practices 📘 Implementation Guide - Comprehensive documentation for all files and components 📝 Detailed Tasks - Complete workflow from project initiation to completion 🔍 Vector-Based Search - Semantic search across your codebase and conversations 🧩 Code Indexing - Automatic detection and indexing of code structures 🔎 Semantic Code Retrieval - Find related code by meaning rather than exact matches 🤖 Automatic Code Analysis - Extract functions, classes, and variables with context

Generate complete project blueprints with file architecture, implementation guides, and full task sequences along with the entire Cursor10x system already implemented!

Overview

The Cursor10x Memory System creates a persistent memory layer for AI assistants (specifically Claude), enabling them to retain and recall:

• Recent messages and conversation history
• Active files currently being worked on
• Important project milestones and decisions
• Technical requirements and specifications
• Chronological sequences of actions and events (episodes)
• Code snippets and structures from your codebase
• Semantically similar content based on vector embeddings
• Related code fragments through semantic similarity
• File structures with function and variable relationships

This memory system bridges the gap between stateless AI interactions and continuous development workflows, allowing for more productive and contextually aware assistance.

System Architecture

The memory system is built on four core components:

1. MCP Server: Implements the Model Context Protocol to register tools and process requests
2. Memory Database: Uses Turso database for persistent storage across sessions
3. Memory Subsystems: Organizes memory into specialized systems with distinct purposes
4. Vector Embeddings: Transforms text and code into numerical representations for semantic search

Memory Types

The system implements four complementary memory types:

1. Short-Term Memory (STM)
   • Stores recent messages and active files
   • Provides immediate context for current interactions
   • Automatically prioritizes by recency and importance
2. Long-Term Memory (LTM)
   • Stores permanent project information like milestones and decisions
   • Maintains architectural and design context
   • Preserves high-importance information indefinitely
3. Episodic Memory
   • Records chronological sequences of events
   • Maintains causal relationships between actions
   • Provides temporal context for project history
4. Semantic Memory
   • Stores vector embeddings of messages, files, and code snippets
   • Enables retrieval of content based on semantic similarity
   • Automatically indexes code structures for contextual retrieval
   • Tracks relationships between code components
   • Provides similarity-based search across the codebase

Features

• Persistent Context: Maintains conversation and project context across multiple sessions
• Importance-Based Storage: Prioritizes information based on configurable importance levels
• Multi-Dimensional Memory: Combines short-term, long-term, episodic, and semantic memory systems
• Comprehensive Retrieval: Provides unified context from all memory subsystems
• Health Monitoring: Includes built-in diagnostics and status reporting
• Banner Generation: Creates informative context banners for conversation starts
• Database Persistence: Stores all memory data in Turso database with automatic schema creation
• Vector Embeddings: Creates numerical representations of text and code for similarity search
• Advanced Vector Storage: Utilizes Turso's F32_BLOB and vector functions for efficient embedding storage
• ANN Search: Supports Approximate Nearest Neighbor search for fast similarity matching
• Code Indexing: Automatically detects and indexes code structures (functions, classes, variables)
• Semantic Search: Finds related content based on meaning rather than exact text matches
• Relevance Scoring: Ranks context items by relevance to the current query
• Code Structure Detection: Identifies and extracts code components across multiple languages
• Auto-Embedding Generation: Automatically creates vector embeddings for indexed content
• Cross-Reference Retrieval: Finds related code across different files and components

Installation

Prerequisites

• Node.js 18 or higher
• npm or yarn package manager
• Turso database account

Setup Steps

1. Configure Turso Database:

Or you can visit Turso and sign up and proceed to create the database and get proper credentials. The free plan will more than cover your project memory.

2. Configure Cursor MCP:

Update .cursor/mcp.json in your project directory with the database url and turso auth token:

Tool Documentation

System Tools

mcp_cursor10x_initConversation

Initializes a conversation by storing the user message, generating a banner, and retrieving context in one operation. This unified tool replaces the need for separate generateBanner, getComprehensiveContext, and storeUserMessage calls at the beginning of each conversation.

Parameters:

• content (string, required): Content of the user message
• importance (string, optional): Importance level ("low", "medium", "high", "critical"), defaults to "low"
• metadata (object, optional): Additional metadata for the message

Returns:

• Object with two sections:
  • display: Contains the banner to be shown to the user
  • internal: Contains the comprehensive context for the agent's use

Example:

mcp_cursor10x_endConversation

Ends a conversation by combining multiple operations in one call: storing the assistant's final message, recording a milestone for what was accomplished, and logging an episode in the episodic memory. This unified tool replaces the need for separate storeAssistantMessage, storeMilestone, and recordEpisode calls at the end of each conversation.

Parameters:

• content (string, required): Content of the assistant's final message
• milestone_title (string, required): Title of the milestone to record
• milestone_description (string, required): Description of what was accomplished
• importance (string, optional): Importance level ("low", "medium", "high", "critical"), defaults to "medium"
• metadata (object, optional): Additional metadata for all records

Returns:

• Object with status and results of each operation

Example:

mcp_cursor10x_checkHealth

Checks the health of the memory system and its database connection.

Parameters:

• None required

Returns:

• Object with health status and diagnostics

Example:

mcp_cursor10x_getMemoryStats

Retrieves detailed statistics about the memory system.

Parameters:

• None required

Returns:

• Object with comprehensive memory statistics

Example:

mcp_cursor10x_getComprehensiveContext

Retrieves a unified context from all memory subsystems, combining short-term, long-term, and episodic memory.

Parameters:

• None required

Returns:

• Object with consolidated context from all memory systems

Example:

Short-Term Memory Tools

mcp_cursor10x_storeUserMessage

Stores a user message in the short-term memory system.

Parameters:

• content (string, required): Content of the message
• importance (string, optional): Importance level ("low", "medium", "high", "critical"), defaults to "low"
• metadata (object, optional): Additional metadata for the message

Returns:

• Object with status and timestamp

Example:

mcp_cursor10x_storeAssistantMessage

Stores an assistant message in the short-term memory system.

Parameters:

• content (string, required): Content of the message
• importance (string, optional): Importance level ("low", "medium", "high", "critical"), defaults to "low"
• metadata (object, optional): Additional metadata for the message

Returns:

• Object with status and timestamp

Example:

mcp_cursor10x_trackActiveFile

Tracks an active file being accessed or modified by the user.

Parameters:

• filename (string, required): Path to the file being tracked
• action (string, required): Action performed on the file (open, edit, close, etc.)
• metadata (object, optional): Additional metadata for the tracking event

Returns:

• Object with status, filename, action and timestamp

Example:

mcp_cursor10x_getRecentMessages

Retrieves recent messages from the short-term memory.

Parameters:

• limit (number, optional): Maximum number of messages to retrieve, defaults to 10
• importance (string, optional): Filter by importance level

Returns:

• Object with status and array of messages

Example:

mcp_cursor10x_getActiveFiles

Retrieves active files from the short-term memory.

Parameters:

• limit (number, optional): Maximum number of files to retrieve, defaults to 10

Returns:

• Object with status and array of active files

Example:

Long-Term Memory Tools

mcp_cursor10x_storeMilestone

Stores a project milestone in the long-term memory.

Parameters:

• title (string, required): Title of the milestone
• description (string, required): Description of the milestone
• importance (string, optional): Importance level, defaults to "medium"
• metadata (object, optional): Additional metadata for the milestone

Returns:

• Object with status, title, and timestamp

Example:

mcp_cursor10x_storeDecision

Stores a project decision in the long-term memory.

Parameters:

• title (string, required): Title of the decision
• content (string, required): Content of the decision
• reasoning (string, optional): Reasoning behind the decision
• importance (string, optional): Importance level, defaults to "medium"
• metadata (object, optional): Additional metadata for the decision

Returns:

• Object with status, title, and timestamp

Example:

mcp_cursor10x_storeRequirement

Stores a project requirement in the long-term memory.

Parameters:

• title (string, required): Title of the requirement
• content (string, required): Content of the requirement
• importance (string, optional): Importance level, defaults to "medium"
• metadata (object, optional): Additional metadata for the requirement

Returns:

• Object with status, title, and timestamp

Example:

Episodic Memory Tools

mcp_cursor10x_recordEpisode

Records an episode (action) in the episodic memory.

Parameters:

• actor (string, required): Actor performing the action (user, assistant, system)
• action (string, required): Type of action performed
• content (string, required): Content or details of the action
• importance (string, optional): Importance level, defaults to "low"
• context (string, optional): Context for the episode

Returns:

• Object with status, actor, action, and timestamp

Example:

mcp_cursor10x_getRecentEpisodes

Retrieves recent episodes from the episodic memory.

Parameters:

• limit (number, optional): Maximum number of episodes to retrieve, defaults to 10
• context (string, optional): Filter by context

Returns:

• Object with status and array of episodes

Example:

Vector-Based Memory Tools

mcp_cursor10x_manageVector

Unified tool for managing vector embeddings with operations for store, search, update, and delete.

Parameters:

• operation (string, required): Operation to perform ("store", "search", "update", "delete")
• contentId (number, optional): ID of the content this vector represents (for store, update, delete)
• contentType (string, optional): Type of content ("message", "file", "snippet", etc.)
• vector (array, optional): Vector data as array of numbers (for store, update) or query vector (for search)
• vectorId (number, optional): ID of the vector to update or delete
• limit (number, optional): Maximum number of results for search operation, defaults to 10
• threshold (number, optional): Similarity threshold for search operation, defaults to 0.7
• metadata (object, optional): Additional info about the vector

Returns:

• Object with status and operation results

Example:

Database Schema

The memory system automatically creates and maintains the following database tables:

• messages: Stores user and assistant messages
  • id: Unique identifier
  • timestamp: Creation timestamp
  • role: Message role (user/assistant)
  • content: Message content
  • importance: Importance level
  • archived: Whether the message is archived
• active_files: Tracks file activity
  • id: Unique identifier
  • filename: Path to the file
  • action: Last action performed
  • last_accessed: Timestamp of last access
• milestones: Records project milestones
  • id: Unique identifier
  • title: Milestone title
  • description: Detailed description
  • timestamp: Creation timestamp
  • importance: Importance level
• decisions: Stores project decisions
  • id: Unique identifier
  • title: Decision title
  • content: Decision content
  • reasoning: Decision reasoning
  • timestamp: Creation timestamp
  • importance: Importance level
• requirements: Maintains project requirements
  • id: Unique identifier
  • title: Requirement title
  • content: Requirement content
  • timestamp: Creation timestamp
  • importance: Importance level
• episodes: Chronicles actions and events
  • id: Unique identifier
  • timestamp: Creation timestamp
  • actor: Actor performing the action
  • action: Type of action
  • content: Action details
  • importance: Importance level
  • context: Action context
• vectors: Stores vector embeddings for semantic search
  • id: Unique identifier
  • content_id: ID of the referenced content
  • content_type: Type of content (message, file, snippet)
  • vector: Binary representation of the embedding vector
  • metadata: Additional metadata for the vector
• code_files: Tracks indexed code files
  • id: Unique identifier
  • file_path: Path to the file
  • language: Programming language
  • last_indexed: Timestamp of last indexing
  • metadata: Additional file metadata
• code_snippets: Stores extracted code structures
  • id: Unique identifier
  • file_id: Reference to the parent file
  • start_line: Starting line number
  • end_line: Ending line number
  • symbol_type: Type of code structure (function, class, variable)
  • content: The code snippet content

Example Workflows

Optimized Conversation Start

Starting a New Session (Alternative Method)

Tracking User Activity

Troubleshooting

Common Issues

1. Database Connection Problems
   • Verify your Turso database URL and authentication token are correct
   • Check network connectivity to the Turso service
   • Verify firewall settings allow the connection
2. Missing Data
   • Check that data was stored with appropriate importance level
   • Verify the retrieval query parameters (limit, filters)
   • Check the database health with mcp_cursor10x_checkHealth()
3. Performance Issues
   • Monitor memory statistics with mcp_cursor10x_getMemoryStats()
   • Consider archiving old data if database grows too large
   • Optimize retrieval by using more specific filters

Diagnostic Steps

1. Check system health:
   const health = await mcp_cursor10x_checkHealth({}); console.log("System Health:", health);
2. Verify memory statistics:
   const stats = await mcp_cursor10x_getMemoryStats({}); console.log("Memory Stats:", stats);
3. Generate a status banner:
   const banner = await mcp_cursor10x_generateBanner({}); console.log("Memory Banner:", banner);

Importance Levels

When storing items in memory, use appropriate importance levels:

• low: General information, routine operations, everyday conversations
• medium: Useful context, standard work items, regular features
• high: Critical decisions, major features, important architecture elements
• critical: Core architecture, security concerns, data integrity issues

License

MIT