Supermemory + Claude Duet
Connect Supermemory to Claude Duet
Prepare a grounded source-backed memory briefing
Read documents with Claude Duet and Agent200. Prepare a source-backed memory briefing within the configured Supermemory account scope.
On this page
Overview
Request an on-demand source-backed memory briefing in Claude Duet. Ask it to name retrieved sources and missing information. This request does not establish continuous monitoring.
Start with one identified document. Preserve returned document content, source and container tags; report missing data and keep the workflow read-only.
Provider reference: Supermemory API.
Use the provider credential described on the Supermemory parent page. Keep it in secure configuration and verify the selected tenant or account before reading.
- Connector
- Supermemory
- Works with
- Claude Duet · Agent app, through MCP
- Authentication
- Your Agent200 API key
- You need
- An Agent200 API key
- Tools
- 4 tools
- MCP server
https://api.agent200.com/mcp/
Before you start
- An Agent200 API key. Create one on the API keys page of the dashboard.
- Supermemory enabled for your project. In the dashboard's Connectors settings, with the endpoints and scopes your agent needs.
- Claude Duet, with custom MCP servers allowed. Claude Duet is a research preview for Pro, Max and Ultra users in the US, by invitation. It uses the connectors of your Claude account, and sending an API key needs Claude's Request headers option.
Connect Supermemory to Claude Duet
-
1
Step 1: Create your Agent200 API key
Open the API keys page in the Agent200 dashboard and choose Create an API key. You add it to Claude Duet in the next step, as "Authorization: Bearer <your key>". One key covers Supermemory and every other connector enabled for your project. Keep it out of chats and source control.
-
2
Step 2: Add Agent200 as a custom connector in Claude
Duet has no connector settings of its own: it uses those of your Claude account. In Claude, go to Customize > Connectors, choose Add > Add custom connector, enter the server URL, choose No sign-in and add the
authorizationrequest header withBearerand your key.Add custom connectorMCP: Claude custom connector (URL and request header) Remote MCP server URL: https://api.agent200.com/mcp/ Authentication: No sign-in Request header name: authorization Value: Bearer YOUR_AGENT200_API_KEY -
3
Step 3: Use it from Duet
Open Claude Duet with the same Claude account. The connectors set up in Claude are there from your first message; manage them in your Claude settings.
-
4
Step 4: Approve actions when Duet asks
Duet asks before it sends a message, pays, signs or deletes anything. If you cannot approve right away, it finishes the rest and brings the approval back later.
Test your connection
Run one request to check everything is wired up.
Ask
Read the Supermemory document I identify and prepare a source-backed memory briefing. Name the source and use only document content, source and container tags. State access gaps; do not change records.
What should come back
- The selected document is identified.
- Only the returned document content, source and container tags are reported.
- Access gaps and missing context are explicit; no source data is changed.
Check the selected document ID, account scope, read permission and pagination. A semantic match is not proof of complete recall. Name returned sources and container scope, and allow for indexing or processing lag.
Get your API key
Create a key in the Agent200 dashboard. It reaches Supermemory and every other connector enabled for your project.
Instructions for AI agents
Using a coding agent? Paste this brief into it, or give it the address of this page. It lists what to do, in order, to connect Supermemory to Claude Duet.
Goal: connect Supermemory to Claude Duet with Agent200, through its MCP server. Guide: https://www.agent200.com/connectors/supermemory/framework/claude-duet/ 1. The user needs an Agent200 API key. If they have none, send them to https://agent200.com/platform/apikeys/ to choose "Create an API key". They paste it into Claude Duet's settings themselves: never ask for it in chat and never write it into files you share. 2. Make sure Supermemory is enabled for their project in the Agent200 dashboard (Connectors), with the tools the task needs. 3. Add Agent200's MCP server to Claude Duet: URL https://api.agent200.com/mcp/, transport streamable HTTP, header "Authorization: Bearer <the user's Agent200 API key>". Follow the setup steps on the guide page for Claude Duet's exact menus or commands. 4. Check it works with this request: "Read the Supermemory document I identify and prepare a source-backed memory briefing. Name the source and use only document content, source and container tags. State access gaps; do not change records." What should come back: - The selected document is identified. - Only the returned document content, source and container tags are reported. - Access gaps and missing context are explicit; no source data is changed. 5. If a tool is missing or a call is refused, check the project's Connectors settings in the dashboard, then retry.
The whole guide is also available as Markdown, for agents and LLMs: /connectors/supermemory/framework/claude-duet.md
Supermemory tools Claude Duet can call
-
Search documents
Readsupermemory_search_documentsSearch indexed documents within a specified container.
-
Read a document
Readsupermemory_get_documentRetrieve one identified document when available.
-
List documents
Readsupermemory_list_documentsFind indexed documents in the key’s scope.
-
Search memories
Readsupermemory_search_memoriesSearch generated memories with source references.
No tool matches that search.
How it works
-
01
You ask Claude Duet
In your own words, for something in Supermemory.
-
02
Claude Duet picks a tool
It sees the Supermemory tools on the Agent200 MCP server and calls the one the request needs.
-
03
Agent200 runs the call
It reaches Supermemory with your Agent200 API key, within the endpoints your project allows.
-
04
Claude Duet answers
It gets the result back from the MCP server and uses it in its reply or its next step.
Permissions at every level. Your project settings decide which services and endpoints the agent may use.
Troubleshooting
The source cannot be read
Check the selected document ID, account scope, read permission and pagination. A semantic match is not proof of complete recall. Name returned sources and container scope, and allow for indexing or processing lag.
FAQ
How do I verify this result?
Compare the named document and returned document content, source and container tags with the source. Treat missing information as missing, not as a fact.
