Many Worlds MCP: How To
This page explains how to connect to the Many Worlds MCP endpoint and use the core research tools.
Endpoint
- URL:
https://manyworldsresearch.com/mcp - Auth: OAuth bearer token or personal MCP bearer token
Keep tokens secret. Do not paste them into chats, screenshots, or logs.
Authentication options
You can connect in either of these ways:
OAuth connector
- Best for ChatGPT connectors and other MCP clients that support OAuth.
- The server publishes MCP OAuth discovery metadata and supports Dynamic Client Registration.
- ChatGPT can create a connector directly against
https://manyworldsresearch.com/mcp.
Personal MCP token
- Best for local tools, scripts, and clients where you want a static bearer token.
- Tokens are created from your Many Worlds account page.
Option 1: Use OAuth
For ChatGPT or another OAuth-capable MCP client:
- Point the client at
https://manyworldsresearch.com/mcp. - Choose OAuth authentication.
- Let the client discover the OAuth metadata automatically.
- Sign in and approve access when redirected.
Notes:
- The server exposes MCP protected resource metadata and authorization server metadata.
- OAuth is proxied through Many Worlds, which validates the client registration and then completes login via Auth0.
- Access tokens issued through OAuth should be sent as
Authorization: Bearer ...on MCP requests.
Option 2: Create a personal MCP token
- Sign in to your Many Worlds account.
- Open Account: https://manyworldsresearch.com/account
- In MCP token, choose a token lifetime (TTL) and create a new token.
- Copy the token immediately and store it in your secret manager.
Notes:
- The full token is shown once at creation time.
- Creating a new token revokes your previous token.
- You can revoke the current token anytime from the Account page.
Use it in requests as:
Authorization: Bearer YOUR_TOKEN
What you can do
The MCP endpoint supports:
list_research— list simulation research instancessearch_research— find research by team/company/event textlist_research_documents— list available documents for one instanceget_research_document— fetch one document bydocument_id
Common document IDs:
metadata_jsonodds_jsonreport_htmlmesh_json
Recommended workflow
- Use
search_researchorlist_researchto find an instance. - Call
list_research_documentsfor that instance. - Fetch only the document you need with
get_research_document. - Start with
metadata_jsonfor quick summaries; usereport_htmlandmesh_jsonfor deep analysis.
Download skill file
If your browser opens it instead of downloading, right-click and choose Save link as....