cd /news/ai-agents/mcp-memory-server-what-it-is-and-how… · home › topics › ai-agents › article
[ARTICLE · art-174269] src=nestdaddy.com ↗ pub= topic=ai-agents verified=true sentiment=↑ positive

MCP memory server: what it is, and how to connect one to Claude, ChatGPT and Cursor

NestDaddy Memory, a hosted MCP memory server at https://nestdaddy.com/memory/mcp, gives AI apps persistent memory through the Model Context Protocol, with setup instructions for Claude, ChatGPT, Cursor, VS Code, Windsurf and Claude Code. The server uses OAuth sign-in instead of an API key and exposes tools including recall, save_memory, search_memories, memory_timeline, log_turn and merge_memories. NestDaddy offers the first month free with no card required.

key takeaway

A hosted MCP memory server is added to Claude Code with `claude mcp add --transport http nestdaddy-memory https://nestdaddy.com/memory/mcp` and OAuth sign-in, and a standing instruction telling the model to call recall on every message and save_memory on decisions makes the AI use it consistently.

read3 min views1 publishedNov 22, 2026
MCP memory server: what it is, and how to connect one to Claude, ChatGPT and Cursor
Image: Nestdaddy (auto-discovered)

An MCP memory server gives any AI app a long-term memory it can read and write through the Model Context Protocol. What it does, local vs hosted, what to look for, and how to connect a hosted one to Claude, ChatGPT, Cursor and Claude Code in two minutes.

The Model Context Protocol (MCP) lets an AI app use outside tools: search the web, read files, call APIs. A memory server is one of the most useful of those tools. It gives the AI a place to save what it learns about you and your work, and to look it up again in any later conversation.

What an MCP memory server does #

It exposes tools like these to the AI:

  • save a fact, a decision or a preference,
  • search for what is relevant to the current question,
  • update a memory when something changes,
  • delete what is no longer true.

The AI decides when to call them. You talk normally; the memory fills up in the background and comes back when it's needed.

Local or hosted? #

Local memory server Hosted memory server
Where it runs On your computer On the provider's servers
Setup Install a package, edit a config file Add one URL, sign in
Works on your phone No Yes
Works in ChatGPT and Claude on the web Usually no Yes
Shared across devices Only if you sync it yourself Automatically
Backups Up to you Handled for you

A local server is great for privacy-first setups on one machine. A hosted server is the easy way to have one memory in every app on every device, including the web and mobile apps that can't run local servers.

What to look for #

  • Search by meaning, not just keywords: you will never ask with the same words you saved.
  • Updates that replace, so the AI never acts on an old decision.
  • Time, so "what did I do last week?" has an answer.
  • Conversation history, so "continue yesterday's work" finds the exact moment.
  • Export and delete, because it's your data.

A hosted MCP memory server in two minutes #

NestDaddy Memory is a hosted MCP memory server. The address is the same for every app:

https://nestdaddy.com/memory/mcp

It uses OAuth, so there is no key to copy: the app opens a sign-in page and you click Allow.

Claude Code

claude mcp add --transport http nestdaddy-memory https://nestdaddy.com/memory/mcp

Then /mcp, choose nestdaddy-memory, Authenticate.

Claude (web and desktop): Customize (or Settings), Connectors, Add custom connector, paste the address, Connect, Allow.

ChatGPT: Settings, Apps & Connectors, add the address (turn on Developer mode under Advanced settings if there is no button), sign in, Allow.

Cursor, VS Code, Windsurf: add a remote (HTTP) MCP server with the address; the browser opens for sign-in.

Older clients that only run a command: npx mcp-remote https://nestdaddy.com/memory/mcp

Make the AI use it every time #

AI apps only call a tool when they think they need it. One standing instruction fixes that:

My main memory is the NestDaddy Memory connector. On every message, first call its recall tool with what I said. When we decide something, save it with save_memory. After each answer, log the turn with log_turn.

The tools your AI gets #

recall (what relates to this message, from memories and past chats), save_memory, search_memories, get_memory, memory_timeline (how a decision changed), timeline_events (what happened in a period), log_turn, search_conversations, read_conversation, plus find_duplicates and merge_memories to keep it tidy.

Start today #

First month free, no card: nestdaddy.com/memory. For more on choosing one, see 8 things to check before you trust an AI memory and persistent memory for AI agents.

── more in #ai-agents 4 stories · sorted by recency
── more on @nestdaddy memory 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
→ Live at https://your-agent.zahid.host ✓
Get free account → Pricing
from €0/mo · no card required
LIVE [news/mcp-memory-server-wh…] indexed:0 read:3min 2026-11-22 · —