Connecting your AI (MCP)
After purchase, you receive ① the MCP server URL to connect to and ② a connection code (activation code). You use your AI's standard MCP connection feature.
The purchase screens, your account page, the Tips page and the legal pages are currently in Japanese only. Your AI can talk with you in your own language.
Two things you receive on your page after purchase
Registering with each AI uses the standard MCP connection method that AI provides. If the screens or menus differ, check each AI's official guide below.
ChatGPT: how to connect (MCP)
You can connect with the MCP connection feature built into each AI. Screens may change with updates or plans, so please also check the official guide below.
*What the Knowledge Access Pass gives you is an MCP server URL dedicated to this service and a one-time connection code (activation code) to enter on the authentication screen. Register these two in your AI's MCP connection settings.
Custom MCP servers can be added only on the web version (not in the mobile apps). In a workspace, an admin must allow “Create plugins with MCPs.” Check the official help for supported plans.
- In ChatGPT on the web, open “Plugins.”
- From +, choose “Add custom MCP server” and enter a name and the connection URL shown on your page. Authentication uses OAuth.
- Review the notice and continue; this service's authentication screen opens. Enter the connection code you issued on your page.
- When the tools appear, ask: “Using the connected Physical AI knowledge, give me an overview of this book.”
Register the connection URL you copied from your page in your AI. Enter the connection code (activation code) only on the first authentication screen; do not paste it into the chat or into MCP settings files.
ChatGPT official setup guide (new tab) ↗Claude: how to connect (MCP)
You can connect with the MCP connection feature built into each AI. Screens may change with updates or plans, so please also check the official guide below.
*What the Knowledge Access Pass gives you is an MCP server URL dedicated to this service and a one-time connection code (activation code) to enter on the authentication screen. Register these two in your AI's MCP connection settings.
Available on Free, Pro, Max, Team and Enterprise (Free allows one custom connector). On Team and Enterprise, an admin adds it under “Organization settings → Connectors,” and each member connects with “Connect.”
- In Claude, open “Customize” → “Connectors” and choose “+ Add” → “Add custom connector.”
- Enter a name and the connection URL shown on your page, then press “Continue.”
- Check the authentication settings and add it; this service's authentication screen opens. Enter the connection code you issued on your page.
- Turn on the connector in a chat and ask about the book's knowledge.
Register the connection URL you copied from your page in your AI. Enter the connection code (activation code) only on the first authentication screen; do not paste it into the chat or into MCP settings files.
Claude official setup guide (new tab) ↗Cursor: how to connect (MCP)
You can connect with the MCP connection feature built into each AI. Screens may change with updates or plans, so please also check the official guide below.
*What the Knowledge Access Pass gives you is an MCP server URL dedicated to this service and a one-time connection code (activation code) to enter on the authentication screen. Register these two in your AI's MCP connection settings.
.cursor/mcp.json inside a project applies only to that project. Normally register it in ~/.cursor/mcp.json, shared by all projects. Your purchased access right and connection codes are not tied to a folder.
- Add a name and the connection URL shown on your page ("url") to "mcpServers" in ~/.cursor/mcp.json. Add it without deleting existing entries.
- Check that the server is enabled in Cursor's “Customize” screen. If asked to authenticate, enter the connection code on this service's authentication screen.
- If it won't connect, check for errors in “MCP Logs” in the Output panel.
- Ask about the book's knowledge from the Agent chat.
{
"mcpServers": {
"physical-ai-knowledge": {
"url": "(connection URL shown on your page)"
}
}
}Register the connection URL you copied from your page in your AI. Enter the connection code (activation code) only on the first authentication screen; do not paste it into the chat or into MCP settings files.
Cursor official setup guide (new tab) ↗Codex: how to connect (MCP)
You can connect with the MCP connection feature built into each AI. Screens may change with updates or plans, so please also check the official guide below.
*What the Knowledge Access Pass gives you is an MCP server URL dedicated to this service and a one-time connection code (activation code) to enter on the authentication screen. Register these two in your AI's MCP connection settings.
Settings are shared by the Codex CLI, IDE extension and desktop app (~/.codex/config.toml).
- In a terminal, run codex mcp add physical-ai-knowledge --url (connection URL shown on your page). In the app, go to “Settings” → “MCP servers” → “Add server,” choose “Streamable HTTP” and enter the URL.
- Run codex mcp login physical-ai-knowledge (“Authenticate” in the app) and enter the connection code on this service's authentication screen.
- Ask Codex about the book's knowledge.
Register the connection URL you copied from your page in your AI. Enter the connection code (activation code) only on the first authentication screen; do not paste it into the chat or into MCP settings files.
Codex official setup guide (new tab) ↗Antigravity: how to connect (MCP)
You can connect with the MCP connection feature built into each AI. Screens may change with updates or plans, so please also check the official guide below.
*What the Knowledge Access Pass gives you is an MCP server URL dedicated to this service and a one-time connection code (activation code) to enter on the authentication screen. Register these two in your AI's MCP connection settings.
The settings file is ~/.gemini/config/mcp_config.json (shared by all workspaces). Register the connection URL as "serverUrl" ("url" and "httpUrl" don't work).
- From “…” at the top of the agent panel, open “MCP Servers” and choose “Manage MCP Servers” → “View raw config.”
- Add a name and the connection URL shown on your page ("serverUrl") to "mcpServers" in mcp_config.json and save. Add it without deleting existing entries.
- On the “Customizations” tab in Agent settings, press “Authenticate.” When this service's authentication screen opens in your browser, enter the connection code you issued on your page. Paste the code shown after authentication into Antigravity and press “Submit.”
- Ask about the book's knowledge from the agent chat.
{
"mcpServers": {
"physical-ai-knowledge": {
"serverUrl": "(connection URL shown on your page)"
}
}
}Register the connection URL you copied from your page in your AI. Enter the connection code (activation code) only on the first authentication screen; do not paste it into the chat or into MCP settings files.
Antigravity official setup guide (new tab) ↗Gemini CLI: how to connect (MCP)
You can connect with the MCP connection feature built into each AI. Screens may change with updates or plans, so please also check the official guide below.
*What the Knowledge Access Pass gives you is an MCP server URL dedicated to this service and a one-time connection code (activation code) to enter on the authentication screen. Register these two in your AI's MCP connection settings.
Listed as reference steps because the operator has not yet verified them. Settings are saved in settings.json (~/.gemini/settings.json to use in all folders, or .gemini/settings.json inside a project to use only there). Authentication requires an environment that can open a browser.
- In a terminal, run gemini mcp add --transport http physical-ai-knowledge (connection URL shown on your page). If you write the settings file directly, register the connection URL as "httpUrl".
- Start Gemini CLI and run /mcp auth physical-ai-knowledge. When this service's authentication screen opens in your browser, enter the connection code you issued on your page.
- Ask about the book's knowledge from the Gemini CLI chat.
{
"mcpServers": {
"physical-ai-knowledge": {
"httpUrl": "(connection URL shown on your page)"
}
}
}Register the connection URL you copied from your page in your AI. Enter the connection code (activation code) only on the first authentication screen; do not paste it into the chat or into MCP settings files.
Gemini CLI official setup guide (new tab) ↗Once connected, try asking
In a new chat, turn on this connector (Physical AI knowledge) before asking. Starting with the big picture helps you grasp how the book is structured.
- Using the connected Physical AI knowledge, give me an overview of this book.
- How do “capability,” “authority” and “delegation” in Chapter 6 differ?
- Rather than which year Physical AI spreads, under what conditions can it be implemented?
- How do “vector,” “goal” and “backcasting” connect?
- Which chapters seem relevant to my work (e.g., planning in manufacturing), and why?
Your AI can show the book's chapters and sections it drew on. Keep asking from the chapters that catch your interest.
Connection flow
You need a plan and permissions that let you set up a custom MCP connection (connector) in your AI. Please check before buying.
Registering an MCP server is a standard feature of each AI. If menu names change, check each AI's official guide.
If you can't connect
Check whether MCP connections are available on the AI side (plan and permissions) and whether authentication has finished. A used connection code cannot be reused for another new connection. If it's connected but the book's information doesn't come up, ask your AI explicitly “using the connected Physical AI knowledge.”