map.name
Your notes, screenshots, links and pages live together in your map.name library. Connect an assistant you already use to find what you saved, ask about it, and keep useful answers.
Once connected, ask your assistant:
Talk to my map.name. Throw things into your own library, or ask it what you have kept.
Talk to someone else's map.name. Read what they have made public, or throw them something to keep.
Claude connects by signing in to map.name. Choose Sign in now and, if Claude shows an OAuth client choice, Register automatically. You can then say put this on my map.name, or ask my map.name, right in the conversation. Free Claude accounts can add one custom connector; Request headers is a separate organization beta and is not needed for this setup.
The coding tools use a separate key-based setup: Claude Code, Codex, Gemini CLI, Antigravity, or any tool that takes a key or speaks MCP. Gemini CLI can also sign in instead of taking a key: add the address, then type /mcp auth mapname.
ChatGPT and the Gemini app take the address and sign in instead of a key: map.name opens, you choose what each may read and whether it may throw into your library, and approve. In ChatGPT you add it yourself at chatgpt.com/plugins: Add, then Create custom MCP server, with the address https://map.name/mcp and OAuth. It stays private to your account. In the Gemini app it is Settings, then Connected Apps, then Add a custom app, which Google offers only to people 18 or over in the United States with a personal account. The Grok app takes the address at grok.com/connectors, under New Connector, then Custom; xAI does not say whether it signs in, so count on it reading what is public. Without sign-in they read what people have made public, your own public pages included. From your own code, the xAI API takes a key, so Grok called that way reaches your whole library.
Throw into your own library, or throw to anyone's address the way you would mail them. What you send to someone else arrives for them to keep or forget.
Everything is private by default, your key is scoped to only what you allow and you can revoke it in one tap, and anything a tool reads back is treated as data, never an instruction to act on.
Ask your map.name in plain language and it answers from what you have kept: it searches across your library and reads any item back, within what your key allows.
Choose your tool above and press Make my key. The page writes your key into what you copy and shows exactly where it goes. You choose how long a key lasts and what it may do, and you can end it at any time from Agents (MCP) in your account menu.
Each tab separates the date its documentation was checked from an actual connection test. Claude and ChatGPT documentation was checked on 9 October 2026. The Claude OAuth picture is an illustrated example; that setup has not yet been tested end to end. ChatGPT's menu and form use supplied references from 8 October, and sign-in was last checked on ChatGPT Pro on 4 October. OpenAI's current setup instructions does not state a plan-by-plan requirement. Other tabs keep their earlier documentation dates, with no new connection test implied.
Claude: support.claude.com/en/articles/11175166 and claude.com/docs/connectors/custom/add-unlisted, checked on 9 October 2026. Claude Code: code.claude.com/docs/en/mcp. ChatGPT: developers.openai.com/api/docs/guides/custom-mcp-server and developers.openai.com/plugins/build/app-quickstart, checked on 9 October 2026; sign-in was last checked in ChatGPT itself on 4 October 2026. Codex: learn.chatgpt.com/docs/extend/mcp. Gemini CLI: geminicli.com/docs/tools/mcp-server, read again on 4 October 2026. The Gemini app: support.google.com/gemini/answer/17209137, read again on 4 October 2026. Antigravity: antigravity.google/docs/mcp. Grok: docs.x.ai/grok/connectors for the app and docs.x.ai/developers/tools/remote-mcp for the xAI API, both read again on 4 October 2026.
Using map.name with your agents is free: connecting them, throwing, searching and reading. No card is asked for.