Memory MCP Server

1# 🧠 Memory MCP Server
2 
3> **Persistent memory and context window caching for LLM conversations.** Save, retrieve, and manage memories with intelligent context archiving. MongoDB-backed storage.
4 
5An [MCP (Model Context Protocol)](https://modelcontextprotocol.io) server that provides memory management and context window caching for AI coding environments like Cursor and Claude Desktop.
6 
7[![Trust Score](https://archestra.ai/mcp-catalog/api/badge/quality/JamesANZ/memory-mcp)](https://archestra.ai/mcp-catalog/jamesanz__memory-mcp)
8 
9## Why Use Memory MCP?
10 
11- 💾 **Persistent Storage** – MongoDB-backed memory that survives sessions
12- 🧠 **Context Caching** – Intelligent archiving and retrieval of conversation context
13- 🏷️ **Tag-based Search** – Organize and find memories by tags
14- 📊 **Relevance Scoring** – Automatically score archived content relevance
15- ⚡ **Easy Setup** – One-click install in Cursor or simple manual setup
16 
17## Quick Start
18 
19Ready to add memory to your AI workflow? Install in seconds:
20 
21**Install in Cursor (Recommended):**
22 
23[🔗 Install in Cursor](cursor://anysphere.cursor-deeplink/mcp/install?name=memory-mcp&config=eyJtZW1vcnktbWNwIjp7ImNvbW1hbmQiOiJucHgiLCJhcmdzIjpbIi15IiwiQGphbWVzYW56L21lbW9yeS1tY3AiXX19)
24 
25**Or install manually:**
26 
27```bash
28npm install -g @jamesanz/memory-mcp
29# Or from source:
30git clone https://github.com/JamesANZ/memory-mcp.git
31cd memory-mcp && npm install && npm run build
32```
33 
34## Features
35 
36### Basic Memory Tools
37- **`save-memories`** – Save memories to database (overwrites existing)
38- **`get-memories`** – Retrieve all stored memories
39- **`add-memories`** – Append new memories without overwriting
40- **`clear-memories`** – Remove all stored memories
41 
42### Context Window Caching
43- **`archive-context`** – Archive conversation context with tags
44- **`retrieve-context`** – Retrieve relevant archived context
45- **`score-relevance`** – Score archived content relevance
46- **`create-summary`** – Create summaries of archived content
47- **`get-conversation-summaries`** – Get all summaries for a conversation
48- **`search-context-by-tags`** – Search archived content by tags
49 
50## Installation
51 
52### Cursor (One-Click)
53 
54Click the install link above or use:
55 
56```
57cursor://anysphere.cursor-deeplink/mcp/install?name=memory-mcp&config=eyJtZW1vcnktbWNwIjp7ImNvbW1hbmQiOiJucHgiLCJhcmdzIjpbIi15IiwiQGphbWVzYW56L21lbW9yeS1tY3AiXX19
58```
59 
60### Manual Installation
61 
62**Requirements:** Node.js 18+, npm, MongoDB
63 
64```bash
65# Clone and build
66git clone https://github.com/JamesANZ/memory-mcp.git
67cd memory-mcp
68npm install
69npm run build
70 
71# Set MongoDB connection string
72export MONGODB_URI="mongodb://localhost:27017"
73 
74# Run server
75npm start
76```
77 
78### Claude Desktop
79 
80Add to `claude_desktop_config.json`:
81 
82**macOS**: `~/Library/Application Support/Claude/claude_desktop_config.json`  
83**Windows**: `%APPDATA%\Claude\claude_desktop_config.json`
84 
85```json
86{
87  "mcpServers": {
88    "memory-mcp": {
89      "command": "node",
90      "args": ["/absolute/path/to/memory-mcp/build/index.js"],
91      "env": {
92        "MONGODB_URI": "mongodb://localhost:27017"
93      }
94    }
95  }
96}
97```
98 
99Restart Claude Desktop after configuration.
100 
101## Configuration
102 
103Set the MongoDB connection string:
104 
105```bash
106export MONGODB_URI="mongodb://localhost:27017"
107```
108 
109Default: `mongodb://localhost:27017`
110 
111## Usage Examples
112 
113### Save Memories
114Store memories from a conversation:
115 
116```json
117{
118  "tool": "save-memories",
119  "arguments": {
120    "memories": ["User prefers TypeScript", "User works on blockchain projects"],
121    "llm": "claude",
122    "userId": "user123"
123  }
124}
125```
126 
127### Retrieve Memories
128Get all stored memories:
129 
130```json
131{
132  "tool": "get-memories",
133  "arguments": {}
134}
135```
136 
137### Archive Context
138Archive conversation context when it gets too long:
139 
140```json
141{
142  "tool": "archive-context",
143  "arguments": {
144    "conversationId": "conv-123",
145    "contextMessages": ["Message 1", "Message 2"],
146    "tags": ["coding", "typescript"],
147    "llm": "claude"
148  }
149}
150```
151 
152### Retrieve Relevant Context
153Get archived content relevant to current conversation:
154 
155```json
156{
157  "tool": "retrieve-context",
158  "arguments": {
159    "conversationId": "conv-123",
160    "tags": ["coding"],
161    "minRelevanceScore": 0.5,
162    "limit": 10
163  }
164}
165```
166 
167## Context Window Caching
168 
169The system automatically:
170- **Archives content** when context usage reaches 80%
171- **Retrieves relevant content** when usage drops below 30%
172- **Scores relevance** using keyword overlap
173- **Creates summaries** to condense long conversations
174 
175## Use Cases
176 
177- **Long Conversations** – Manage context windows for extended sessions
178- **Memory Persistence** – Save important information across sessions
179- **Context Retrieval** – Bring back relevant past conversations
180- **Research Projects** – Organize and tag research conversations
181 
182## Technical Details
183 
184**Built with:** Node.js, TypeScript, MCP SDK, MongoDB  
185**Dependencies:** `@modelcontextprotocol/sdk`, `mongodb`, `zod`  
186**Platforms:** macOS, Windows, Linux
187 
188**Storage:** MongoDB (default: `mongodb://localhost:27017`)
189 
190## Contributing
191 
192⭐ **If this project helps you, please star it on GitHub!** ⭐
193 
194Contributions welcome! Please open an issue or submit a pull request.
195 
196## License
197 
198ISC
199 
200## Support
201 
202If you find this project useful, consider supporting it:
203 
204**⚡ Lightning Network**
205```
206lnbc1pjhhsqepp5mjgwnvg0z53shm22hfe9us289lnaqkwv8rn2s0rtekg5vvj56xnqdqqcqzzsxqyz5vqsp5gu6vh9hyp94c7t3tkpqrp2r059t4vrw7ps78a4n0a2u52678c7yq9qyyssq7zcferywka50wcy75skjfrdrk930cuyx24rg55cwfuzxs49rc9c53mpz6zug5y2544pt8y9jflnq0ltlha26ed846jh0y7n4gm8jd3qqaautqa
207```
208 
209**₿ Bitcoin**: [bc1ptzvr93pn959xq4et6sqzpfnkk2args22ewv5u2th4ps7hshfaqrshe0xtp](https://mempool.space/address/bc1ptzvr93pn959xq4et6sqzpfnkk2args22ewv5u2th4ps7hshfaqrshe0xtp)
210 
211**Ξ Ethereum/EVM**: [0x42ea529282DDE0AA87B42d9E83316eb23FE62c3f](https://etherscan.io/address/0x42ea529282DDE0AA87B42d9E83316eb23FE62c3f)
212