# How to Connect Claude, ChatGPT, Codex and Cursor to LotsNotes

Source: https://lotsnotes.com/blog/connect-claude-chatgpt-codex-cursor-to-lotsnotes
Published: 2026-10-13T06:36:08.846+00:00

To connect Claude, ChatGPT, Codex or Cursor to LotsNotes, add one MCP server URL, `https://api.lotsnotes.com/mcp`, to the agent and sign in with your Lots account. There's no key to paste: Claude and ChatGPT add it as a custom connector or app, Claude Code and Codex take one terminal command, and Cursor takes three lines of JSON. Cloud agents that can't use MCP read the same notes over the REST API with a key.

I run every LotsTech agent this way. Their instructions, plans, handoffs and ledgers live in one LotsNotes workspace, and Claude Code, Codex and cloud agents read the same notes before they work. Below is the exact setup for each client, checked against each vendor's own docs on 13 October 2026, then a short read-only run from my own session so you can see what the agent gets back.

## What you need

- A LotsNotes account. Free covers one person: 1 workspace, 3 folders, unlimited notes and checklists, and MCP and REST access. No card.
- At least one shared note. Start with a note titled `AGENTS.md` (more on that below) and one note with your positioning or rules.
- The agent you use: Claude, Claude Code, ChatGPT, Codex, Cursor, or a cloud agent that can make HTTP requests.

The server URL is the same everywhere:

```text
https://api.lotsnotes.com/mcp
```

## Write an AGENTS.md note first

Agents read better when you tell them where to start. In LotsNotes, a note titled `AGENTS.md` is the instruction file agents read first.

- **At workspace level**, it comes back with every folder's notes, so house rules reach every agent in every folder.
- **Inside a folder**, it holds the rules for that product, client or agent. My workspace has one `AGENTS.md` at the top and one in each agent's folder (Blog, Social, Research and so on).

Keep it short: who you are, what the agent should read next, and what it must never do. The agent finds notes by title and reads only what the task needs, so the rest of your context stays out of the way until it's useful.

Notes are **shared** or **private**. Shared notes follow your workspace roles. Private notes stay on lotsnotes.com and are never returned to agents or the API: to an agent, a private note looks like it doesn't exist. Keep anything you don't want an agent to read private.

## Claude (claude.ai and Claude Desktop)

1. Open **Customize**, then **Connectors**.
2. Click **+ Add**, then **Add custom connector**.
3. Name it `LotsNotes` and paste `https://api.lotsnotes.com/mcp` as the remote MCP server URL.
4. Keep the detected sign-in settings and click **Add**.
5. Connect and sign in with your Lots account when the LotsNotes page asks you to allow Claude.

To use it in a chat, open the **+** menu at the lower left of the chat box, choose **Connectors**, and switch LotsNotes on.

On Claude Team and Enterprise, an owner adds the custom connector once under **Organization settings**, **Connectors**; each member then finds it under **Customize**, **Connectors** and clicks **Connect** to sign in with their own account. Anthropic's help center says Free plans can add one custom connector.

## Claude Code

Run this in your terminal:

```bash
claude mcp add --transport http lots-notes https://api.lotsnotes.com/mcp
```

Then sign in. Inside Claude Code, run `/mcp`, select **lots-notes** and choose **Authenticate**; your browser opens the LotsNotes sign-in. From the shell, `claude mcp login lots-notes` does the same.

By default the server is saved for the current project only (`local` scope). To have LotsNotes in every project, add `--scope user`:

```bash
claude mcp add --transport http --scope user lots-notes https://api.lotsnotes.com/mcp
```

Use `--scope project` to list it in the repository's `.mcp.json` for your team. Everyone still signs in with their own account and sees only the folders their role allows.

## ChatGPT

ChatGPT connects to remote MCP servers as custom apps in developer mode, on ChatGPT web.

1. Turn on developer mode: **Settings**, **Apps**, **Advanced settings**, **Developer mode**. On Business workspaces, admins and owners turn it on; on Enterprise and Edu, admins grant it to members.
2. Go to **Apps**, then **Create**.
3. Name it `LotsNotes`, paste `https://api.lotsnotes.com/mcp` as the endpoint and choose OAuth.
4. Click **Scan Tools**, complete the LotsNotes sign-in, wait for the scan, then click **Create**.
5. In a new chat, pick LotsNotes from the tools menu or mention it in your prompt.

OpenAI's help center says full MCP, including write actions, is available on ChatGPT Business, Enterprise and Edu, and Pro users can connect MCP servers with read and fetch permissions in developer mode. ChatGPT may ask you to confirm before a write, such as saving a note.

## Codex

Run:

```bash
codex mcp add lots-notes --url https://api.lotsnotes.com/mcp
codex mcp login lots-notes
```

The first command writes the server to `~/.codex/config.toml`:

```toml
[mcp_servers.lots-notes]
url = "https://api.lotsnotes.com/mcp"
```

The second runs the LotsNotes sign-in in your browser. OpenAI's docs say the Codex CLI, the IDE extension and the ChatGPT desktop app share this configuration, so you set it up once. Type `/mcp` in Codex to see your active servers.

## Cursor (and Windsurf and other MCP clients)

Add LotsNotes to `~/.cursor/mcp.json` to use it in every project, or to `.cursor/mcp.json` in one project:

```json
{
  "mcpServers": {
    "lots-notes": {
      "url": "https://api.lotsnotes.com/mcp"
    }
  }
}
```

Cursor supports OAuth for remote servers, so it opens the LotsNotes sign-in the first time it connects. You can switch the server on or off from **Customize** in Cursor's sidebar, and if something looks wrong, the **Output** panel has an **MCP Logs** view. Windsurf and most other clients use the same `mcpServers` shape with a `url`.

## Cloud agents and your own code (REST)

Cloud agents that run on their own servers, and scripts you write, use the REST API instead of MCP.

1. Create a key at [api.lotsnotes.com/dashboard](https://api.lotsnotes.com/dashboard), or in the LotsTech key manager at [lotstech.com/api](https://www.lotstech.com/api) (choose Lots Notes). The key is shown once. Store it as a secret in the agent's environment; don't paste it into a chat.
2. Give the agent the docs: `https://api.lotsnotes.com/docs.md`. It lists all 22 endpoints with examples.
3. The agent calls the API with the key in a header:

```bash
# list your workspaces
curl -H "X-API-Key: $LOTSNOTES_API_KEY" \
  https://api.lotsnotes.com/api/v1/lotsnotes/workspaces

# find AGENTS.md notes in one workspace
curl -H "X-API-Key: $LOTSNOTES_API_KEY" \
  "https://api.lotsnotes.com/api/v1/lotsnotes/context?workspace_id=<workspace_id>&query=AGENTS.md"
```

`Authorization: Bearer <key>` works too. The default limit is 100 requests a minute per key. The same rules apply as over MCP: shared notes within your roles, never private ones.

## What the agent sees: a read-only run

Once you're connected, the agent works the same way in every client. Here's what came back when I ran the LotsNotes tools from my own Claude Code session on 13 October 2026. I only read; nothing was created or changed.

**1. List workspaces.** `lotsnotes_list_workspaces` returned my workspaces with each one's id, name and my role (owner). One of them is **LotsTech Agents**, where every agent's notes live.

**2. Find the AGENTS.md notes.** `lotsnotes_find_context` with that workspace and the query `AGENTS.md` returned:

- the workspace-level `AGENTS.md` (level `workspace`), which comes back on every search in that workspace;
- one `AGENTS.md` per agent folder, each with its `folder_path`, such as `Blog`, `Blog / lots.notes`, `Social`, `Research` and `Product Ops`.

Every result carries a one-line summary and its `https://lotsnotes.com/note/<id>` link, so the agent can choose what to open, and hand you the link to check what it read.

**3. Read the note.** `lotsnotes_read_note` on the workspace `AGENTS.md` returned the note as Markdown with its `current_version` and link. In my case it tells every agent to read that note whole, then its folder's `AGENTS.md`, then its handoff note, and to save its work back to its folder before it stops.

That's the whole loop: list, find by title, read only what the task needs. When the agent saves a change, it sends the version it read, so if someone edited the note in the meantime it gets a conflict instead of overwriting their edit. Every save keeps a version you can restore on lotsnotes.com.

## Your first prompt

Paste this into whichever agent you connected:

> Use LotsNotes. List my workspaces, open the one for this business, and read the workspace AGENTS.md, then the AGENTS.md in the folder for [product or client]. Summarize the rules you'll follow and give me the links to the notes you read. Don't change anything yet.

Then ask for real work: "Read our positioning note before you write this landing page," or "Save this decision to the Product folder and send me the link."

## See and cut off what's connected

The **Workspace** page on lotsnotes.com lists your **Connected apps** with when each was last used. Disconnect one and it loses access everywhere. Agents always act as you, with your role, and see only the folders you can access.

## What it costs

Connecting agents is free. Free covers one person with 1 workspace, 3 folders, unlimited notes and checklists, and MCP and REST access. **LotsNotes Pro is $99, pay once**, for unlimited workspaces, folders and team members with roles. It's already included while you have a Pro or Business plan of LotsSocial, LotsBlog, LotsTeam or LotsCRM. Details are on the [pricing page](https://lotsnotes.com/pricing).

If you're still choosing where your business context should live, I compared ten options in [the knowledge base and notes tools with MCP support](https://lotsnotes.com/blog/best-knowledge-base-tools-with-mcp-support).

## Common questions

### Do I need an API key to connect Claude, ChatGPT, Codex or Cursor?

No. Those clients connect to `https://api.lotsnotes.com/mcp` with OAuth: you sign in with your Lots account in the browser. You need an API key only for the REST API, which cloud agents and your own scripts use.

### Which ChatGPT plans can connect LotsNotes?

ChatGPT connects remote MCP servers as custom apps in developer mode. OpenAI's help center says full MCP, including writes, is on Business, Enterprise and Edu, and Pro users can connect MCP servers with read and fetch permissions. Check OpenAI's page for your plan, since it is in beta and changing.

### Can my AI agent read my private notes?

No. Private notes stay on lotsnotes.com and are never returned through MCP or the API. To an agent, a private note looks like it doesn't exist. Shared notes follow your workspace roles.

### What is an AGENTS.md note?

A note titled `AGENTS.md` holds the instructions agents read first. At workspace level it reaches every folder; inside a folder it holds the rules for that product, client or agent.

### Do I have to write my context again for each agent?

No. That's the point: you write it once in LotsNotes and every connected agent reads the same current notes. Connecting a new agent is just the setup step for that client.

### How do I disconnect an agent?

Open the Workspace page on lotsnotes.com. Connected apps lists each agent with when it was last used; disconnect it there and it loses access everywhere.

### Does LotsNotes work with other agents, like Windsurf or Gemini?

Any client that supports remote MCP servers can use the same URL, and any agent that can make HTTP requests can use the REST API. The [setup page](https://lotsnotes.com/mcp) lists the common ones.

## Connect your first agent

Write an `AGENTS.md` note, add the server to the agent you use most, and ask it to read the note back to you.

```bash
claude mcp add --transport http lots-notes https://api.lotsnotes.com/mcp
```

Setup for every client is on [lotsnotes.com/mcp](https://lotsnotes.com/mcp). Free for one person; Pro is one payment. [See pricing](https://lotsnotes.com/pricing).

*Client steps checked on 13 October 2026 from each vendor's own docs: [Anthropic: custom connectors](https://support.claude.com/en/articles/11175166-getting-started-with-custom-connectors-using-remote-mcp), [Claude Code MCP](https://code.claude.com/docs/en/mcp), [OpenAI: developer mode and MCP apps in ChatGPT](https://help.openai.com/en/articles/12584461-developer-mode-and-mcp-apps-in-chatgpt), [Codex MCP](https://learn.chatgpt.com/docs/extend/mcp?surface=cli), [Cursor MCP](https://cursor.com/docs/context/mcp). LotsNotes details from [lotsnotes.com/mcp](https://lotsnotes.com/mcp), [lotsnotes.com/skill.md](https://lotsnotes.com/skill.md), [api.lotsnotes.com/docs.md](https://api.lotsnotes.com/docs.md) and [lotsnotes.com/pricing](https://lotsnotes.com/pricing). Client menus change; if a step looks different, the vendor's page has the latest.*