claude.ai
Add twin memory to claude.ai as a custom connector. Each person signs in and chooses whose memory Claude reaches, with no key or header to paste.
What you get
Claude searches your twin memory before it answers, and saves what it learns as you talk. It reaches one end user's memory, your own by default: the same memory your app's backend and your other assistants read and write.
The connector lists the eight read tools, and memory_add when you allow saving. Every tool is on the
tools reference.
https://api.twinmemory.dev/v1/mcpPlans that support it
Custom connectors work on every claude.ai plan:
| Plan | Custom connectors |
|---|---|
| Free | One |
| Pro and Max | Yes, added by each person |
| Team and Enterprise | Yes, added by an Owner for the organisation; each member connects as themselves |
A connector you add in claude.ai also appears in Claude Desktop and the mobile apps. Connections come from Anthropic's cloud, not from your device.
Add the connector
On Team and Enterprise an Owner adds it once: see For admins. On Pro and Max:
Open Customize > Connectors (claude.ai/customize/connectors) and select Add custom connector.
Name it twin memory and set the URL to https://api.twinmemory.dev/v1/mcp, exactly, with no trailing slash.
Under Authentication, keep Sign in now. Under OAuth client, choose Use Claude's published identity (recommended). No client ID or secret is needed: twin memory reads Claude's identity from the address it publishes. Register automatically works as well.
Select Add. A twin memory window opens: sign in, or create an account, then choose on the connect card and select Allow. You are returned to claude.ai with the connector on.
Prefer Sign in now to Sign in when needed: the connector is ready at once, and a problem with signing in shows while you are still in the dialog. Authentication settings cannot be changed after a connector is added; to change them, remove it and add it again.
What happens when you sign in
After you sign in, twin memory shows one card naming the app (here, Claude, from claude.ai) and four choices,
already filled in:
| Choice | What it means |
|---|---|
| Organisation | Shown only if you belong to more than one. |
| Project | Which project's memory, as for an API key. |
| End user | Whose memory Claude reaches. Your own by default, made from your email: dana.cole@acme.com is dana.cole. Change it to reach someone else's, as X-Twin-Account would. See End users. |
| Access | Read and save lists memory_add; Read only does not. Viewers in the organisation can grant Read only and nothing more. |
Allow records the connection; Cancel tells Claude you declined. Connect the same app again later with the same choices and the card is skipped. It shows again if the app asks for more than you allowed, or after you disconnect.
Make it save everything
twin memory is meant to hold almost everything a person tells their assistant, so their other apps can use it too. claude.ai does not pass a server's instructions to Claude, and it loads connector tools only when needed, so the guidance travels in two places:
- The
memory_addtool's own description tells Claude to save on its own, every turn, small details included, and never secrets. Nothing to set up. - Your claude.ai preferences make it stick. Paste this into Settings › Profile › personal preferences, or into a Project's instructions to apply it to one project:
Use twin memory as my memory. Before answering anything about me, search it with memory_search. Save everything new I tell you about me, my work, and the people, places and things around me, small details included, with memory_add, on your own and without asking. Never save secrets, passwords or keys.Claude has its own built-in memory as well; the text above tells it which one to use for things about you. To see what was saved, open the console's Playground or End users page for the end user you chose when connecting.
Tools and approvals
Every tool is sent with a title and hints. The read tools are marked read-only, so Claude runs them without
asking. memory_add is marked as a write that never deletes, so Claude asks first: Allow once or Always
allow. Set it per tool in the connector's settings: Always allow, Needs approval or Blocked.
Research mode
Claude's Research mode runs connector tools without asking. If you do not want saves there, set memory_add
to Blocked, or connect with Read only.
Disconnecting
- In twin memory: Account > Connected apps > Disconnect. Owners and Admins can also disconnect anyone's connection on the console's MCP page, under Connections.
- In claude.ai: open the connector under Customize > Connectors and disconnect or remove it.
After a disconnect in twin memory, the token Claude holds works for up to 15 more minutes. Claude then shows the connector as needing sign-in, and connecting again shows the card.
For admins
On Team and Enterprise:
An Owner opens Organization settings > Connectors (claude.ai/admin-settings/connectors), selects Add, then Custom, then Web. On Team only Owners and Primary Owners can; on Enterprise also a custom role that manages connectors.
Set the URL to https://api.twinmemory.dev/v1/mcp and choose Use Claude's published identity.
Members find it under Customize > Connectors, labelled Custom. Each one selects Connect and signs in to twin memory as themselves, so each reaches their own end user, with the access their role allows.
Each member needs a twin memory account in your organisation: invite them under Team in the console first.
Troubleshooting
- Sign-in loops, or "Authorization with the MCP server failed". The URL must be exactly
https://api.twinmemory.dev/v1/mcp. Remove the connector and add it again with the right URL. - Claude reads the wrong end user. Disconnect under Account > Connected apps, then connect again in claude.ai and choose the end user on the card.
- The connector needs sign-in again. You or an admin disconnected it, or your role changed so it no longer allows the connection's access (a Read and save connection after you became a Viewer). Connect again.
memory_addis missing. The connection is Read only, or the tool is Blocked in Claude.- Running twin memory yourself, through a tunnel. Claude needs a public HTTPS address. Set
PUBLIC_URLand follow the README's "Connect claude.ai or ChatGPT from your machine". On ngrok's free plan, open the address in your browser once and pass its warning page before you sign in. - A strict firewall. Claude connects from
160.79.104.0/21. Allow it to reach bothapi.twinmemory.devandtwinmemory.dev, where you sign in.
Sources checked 2026-10-22: claude.com/docs: add a custom connector, claude.com/docs: authentication, claude.com/docs: getting started, support.claude.com: custom connectors using remote MCP.