MCP 服务器

Claude Code Memory Server

Provides persistent memory capabilities for Claude Code using Neo4j graph database to track development tasks, code patterns, solutions, and their relationships across sessions and projects, enabling contextual assistance and pattern recognition.

README

Claude Code Memory Server

A Neo4j-based Model Context Protocol (MCP) server that provides intelligent memory capabilities for Claude Code, enabling persistent knowledge tracking, relationship mapping, and contextual development assistance.

Overview

This MCP server creates a sophisticated memory system that tracks Claude Code's activities, decisions, and learned patterns to provide contextual memory across sessions and projects. It uses Neo4j as a graph database to capture and analyze complex relationships between development concepts, solutions, and workflows.

Features

Core Memory Operations

Persistent Memory Storage - Store development tasks, solutions, and patterns
Intelligent Search - Find relevant memories by context, content, or relationships
Relationship Mapping - Track how different concepts, files, and solutions relate
Context Awareness - Project-specific and technology-specific memory retrieval

Advanced Intelligence

Pattern Recognition - Automatically identify reusable development patterns
Solution Effectiveness - Track and learn from successful approaches
Workflow Memory - Remember and suggest optimal development sequences
Error Prevention - Learn from past mistakes to prevent similar issues

Development Integration

Task Execution Tracking - Monitor what Claude Code does and how
Code Pattern Analysis - Identify and store successful code patterns
Project Context Memory - Understand codebase conventions and dependencies
Collaborative Learning - Share knowledge across development sessions

Architecture

Memory Types

Task - Development tasks and their execution patterns
CodePattern - Reusable code solutions and architectural decisions
Problem - Issues encountered and their context
Solution - How problems were resolved and their effectiveness
Project - Codebase context and project-specific knowledge
Technology - Framework, language, and tool-specific knowledge

Relationship Types

The system tracks seven categories of relationships:

Causal - CAUSES, TRIGGERS, LEADS_TO, PREVENTS, BREAKS
Solution - SOLVES, ADDRESSES, ALTERNATIVE_TO, IMPROVES, REPLACES
Context - OCCURS_IN, APPLIES_TO, WORKS_WITH, REQUIRES, USED_IN
Learning - BUILDS_ON, CONTRADICTS, CONFIRMS, GENERALIZES, SPECIALIZES
Similarity - SIMILAR_TO, VARIANT_OF, RELATED_TO, ANALOGY_TO, OPPOSITE_OF
Workflow - FOLLOWS, DEPENDS_ON, ENABLES, BLOCKS, PARALLEL_TO
Quality - EFFECTIVE_FOR, INEFFECTIVE_FOR, PREFERRED_OVER, DEPRECATED_BY, VALIDATED_BY

Installation

Prerequisites

Python 3.10 or higher
Neo4j database (local or cloud)
Claude Code with MCP support

Setup

Clone the repository:

git clone https://github.com/viralvoodoo/claude-code-memory.git
cd claude-code-memory

Install dependencies:

pip install -e .

Set up Neo4j connection:

cp .env.example .env
# Edit .env with your Neo4j credentials

Initialize the database schema:

python -m claude_memory.setup

Configuration

Environment Variables

NEO4J_URI - Neo4j database URI (default: bolt://localhost:7687)
NEO4J_USER - Database username (default: neo4j)
NEO4J_PASSWORD - Database password
MEMORY_LOG_LEVEL - Logging level (default: INFO)

Claude Code Integration

Add to your Claude Code MCP configuration:

{
  "mcpServers": {
    "claude-memory": {
      "command": "python",
      "args": ["-m", "claude_memory.server"],
      "env": {
        "NEO4J_URI": "bolt://localhost:7687",
        "NEO4J_USER": "neo4j",
        "NEO4J_PASSWORD": "your-password"
      }
    }
  }
}

Usage

Available MCP Tools

Core Memory Operations

store_memory - Store new development memories with context
get_memory - Retrieve specific memory by ID with relationships
search_memories - Find memories by content, context, or relationships
update_memory - Modify existing memory content
delete_memory - Remove memory and cleanup relationships

Relationship Management

create_relationship - Link memories with specific relationship types
get_related_memories - Find memories connected to a specific memory
analyze_relationships - Discover relationship patterns in memory graph

Development Intelligence

analyze_codebase - Scan project and create contextual memory graph
track_task_execution - Record development workflow and patterns
suggest_similar_solutions - Find analogous past solutions
predict_solution_effectiveness - Estimate success probability of approaches

Advanced Analytics

get_memory_graph - Visualize knowledge network and relationships
find_memory_paths - Discover connection chains between concepts
memory_effectiveness - Track and analyze solution success rates

Development

Project Structure

claude-code-memory/
├── src/claude_memory/          # Main source code
│   ├── __init__.py
│   ├── server.py              # MCP server implementation
│   ├── models.py              # Data models and schemas
│   ├── database.py            # Neo4j database operations
│   ├── memory_store.py        # Core memory logic
│   ├── relationships.py       # Relationship management
│   ├── search.py              # Search and retrieval
│   └── intelligence.py        # Pattern recognition and analytics
├── tests/                     # Test suite
├── docs/                      # Documentation
├── scripts/                   # Utility scripts
└── pyproject.toml            # Project configuration

Development Setup

# Install development dependencies
pip install -e ".[dev]"

# Install pre-commit hooks
pre-commit install

# Run tests
pytest

# Format code
black src/ tests/
ruff --fix src/ tests/

# Type checking
mypy src/