How to give Claude or ChatGPT long-term memory with MCP

You can give Claude, ChatGPT or Cursor long-term memory by connecting them to a knowledge base over MCP (Model Context Protocol). Point the client at an MCP endpoint, authenticate, and it can read everything you've saved — across sessions, across tools, across model upgrades. Here's the setup for each client.

How to give Claude or ChatGPT long-term memory using MCP — step by step

BrainTube's MCP server is remote — there is nothing to install. Point your client at the endpoint and authenticate — Claude.ai signs in over OAuth, everything else passes an API key in the X-BrainTube-Token header — and you're done.

The endpoint

Every client below points at the same streamable-HTTP endpoint:

https://mcp.brain-tube.com/mcp

1. Claude.ai — one-click OAuth

Open Claude.ai → Settings → Connectors → Add custom connector. Paste the endpoint above and sign in when prompted. No API key needed; Claude.ai connects via OAuth.

2. Claude Desktop, Cursor and other MCP clients — JSON config

These clients share one config shape — the same server block works in each of them:

{
  "mcpServers": {
    "braintube": {
      "type": "streamable-http",
      "url": "https://mcp.brain-tube.com/mcp",
      "headers": {
        "X-BrainTube-Token": "bt_your_key_here"
      }
    }
  }
}

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json
Cursor: Settings → MCP → Add Server (same URL and header)

3. ChatGPT — in testing

ChatGPT support is in testing — both the Developer-mode connector and the custom-GPT path. Until that's verified, Claude.ai, Claude Desktop and Cursor are the documented routes; any client that speaks MCP streamable-http can try the same endpoint and config.

4. Get your API key

Sign in and open Settings → API Keys to create one. Keys are shown once at creation, hashed at rest, and can be rotated or revoked at any time. Replace bt_your_key_here in the config with your real key and restart the client.

The copy-pasteable version of every config above lives on the connect page.

The pain: every session starts from zero

You explained the whole project to Claude last week. Today, it remembers none of it. So you paste the same context again. Then again in ChatGPT. Then again in Cursor. Every model upgrade resets the relationship.

The problem isn't the AI. It's that your knowledge isn't anywhere the AI can actually read.

What an MCP memory layer is

MCP (Model Context Protocol) is an open standard that lets AI clients pull live context from external sources through a single, well-typed interface. Instead of a vendor's tiny built-in "memory" feature, MCP gives any compliant AI a way to read a real knowledge base you own.

A memory layer over MCP means: one knowledge base, many AIs, no copy-paste.

How BrainTube does it

Auto-capture. BrainTube ingests from YouTube (segmented transcripts), podcasts and audio, PDFs and EPUBs, web articles, Notion (4-hour sync), meeting audio, and screenshots.
Compile. Each item gets semantic embeddings and entity extraction. A Postgres knowledge graph auto-links concepts; PageRank-style retrieval surfaces what matters. FSRS spaced repetition keeps the important bits sharp.
Query from any AI. A live MCP server (listed in the official MCP registry) exposes your library to Claude, Cursor and any MCP-capable client.

The differentiator: memory that fills itself

Almost every "AI memory" or "second brain" tool is a note app you have to type into. BrainTube is the opposite shape: the memory grows from your existing consumption — the videos you already watch, the podcasts you already listen to, the PDFs you already open.

Tools change. Models cycle. Your mind's index stays.

Frequently asked

How do I give Claude or ChatGPT long-term memory using MCP?

Connect the client to an MCP server that exposes your knowledge base. In Claude.ai that is Settings → Connectors → Add custom connector with the endpoint https://mcp.brain-tube.com/mcp. In Claude Desktop, Cursor and other clients you add a streamable-http entry to the client's JSON config with an X-BrainTube-Token header. Once connected, the assistant can read your saved library on demand in every future session.

Can I give Claude or ChatGPT long-term memory?

Yes for Claude today — connect it to a BrainTube knowledge base over MCP and it reads what you've saved across sessions instead of starting from zero every chat. ChatGPT support is in testing.

How is this different from ChatGPT's built-in memory?

Built-in memory is small, vendor-locked, and only inside that one product. A BrainTube knowledge base is yours, queryable from any MCP-compatible AI, and you don't have to type anything in — it fills itself from what you already watch and read.

Do I have to type notes manually?

No. That's the whole wedge. BrainTube auto-captures from YouTube videos, podcasts, PDFs, EPUBs, web articles, Notion, meeting audio, and screenshots. The memory layer grows from your existing consumption.

Which AI tools can connect?

Any MCP-capable client: Claude Desktop and Claude.ai (one-click OAuth), Cursor, plus CLI tools and any future MCP client. ChatGPT support is in testing. BrainTube is listed in the official MCP registry.

Is my saved data private to me?

Each user's corpus is private to their own account. Your library is not shared, pooled, or used to train models on other users' behalf.

Where do I get an API key?

Sign in to BrainTube and create one under Settings → API Keys. Keys are shown once at creation, hashed at rest, and can be rotated or revoked at any time. Each key has its own audit log.

Do I need a paid plan to connect my AI over MCP?

No. MCP access is included on every plan, including Free. What differs by tier is the credit allowance: saving items, running searches and asking your brain all spend credits, and higher tiers get more of them. Free starts with 30 credits a month.