AI 原生开放教材

外观

Claude Code Everything You Need to Know工程手册·英文原文 ↗仓库 ↗

Memory MCP Server ​

Overview ​

The Memory server allows Claude Code to store and retrieve information across sessions, providing persistent context, remembering project details, and maintaining user preferences.

Note: This MCP server is distinct from Claude Code's built-in auto memory (on by default since v2.1.59), which stores notes under ~/.claude/projects/<project>/memory/ with a MEMORY.md index and is managed via the /memory command.

Installation ​

Install Memory globally for cross-project persistence:

bash
claude mcp add memory -s user -- npx -y @modelcontextprotocol/server-memory

Local Installation (Project-Specific) ​

For project-specific memory storage:

bash
claude mcp add memory -s local -- npx -y @modelcontextprotocol/server-memory

Usage ​

Step 2: Using Memory in Claude Code ​

Once installed, the server exposes knowledge-graph tools that Claude calls when you ask it to remember or recall something. Usage is not automatic — to make it so, add an instruction to your CLAUDE.md (e.g., "Use the memory server to store and retrieve project context").

Example: Storing Information ​

"Remember that the current project uses TypeScript"

Claude stores this information for future reference.

Example: Retrieving Information ​

Later in the same or a new session:

"What programming language is used in the current project?"

Claude responds:

"TypeScript"

Features ​

  • Persistent context: Information persists across sessions
  • Project-specific notes: Remember project details, conventions, and preferences
  • User preferences: Store personal preferences and workflow habits
  • On-request memory: Claude calls the server's knowledge-graph tools when you ask it to remember or recall
  • Cross-session memory: Recall information from previous conversations

Use Cases ​

  • Remember project-specific coding conventions
  • Store frequently used commands or configurations
  • Maintain context about ongoing tasks
  • Track decisions made in previous sessions
  • Remember team preferences and guidelines

Integration with Other Servers ​

Memory works seamlessly with other MCP servers:

  • With Serena: Combines persistent memory with semantic code understanding
  • With Sequential Thinking: Enhances reasoning with historical project context
  • With Playwright: Remembers browser automation preferences and configurations

Tip: Global installation enables memory sharing across all projects, while local installation keeps project memories separate.

Verification ​

Verify Memory is connected:

bash
claude mcp list

Expected output:

memory: npx -y @modelcontextprotocol/server-memory - ✓ Connected

Best Practices ​

  • Use global installation for cross-project memory and user preferences
  • Use local installation for project-specific isolated memory
  • Regularly test memory recall to ensure proper functioning
  • Combine with Serena for enhanced project-aware context

Troubleshooting ​

If Memory fails to connect:

  1. Verify Node.js is installed: node --version
  2. Check npx availability: npx --version
  3. Remove and reinstall:
    bash
    claude mcp remove memory -s user
claude mcp add memory -s user -- npx -y @modelcontextprotocol/server-memory

For more help, see the Troubleshooting Guide.