Connecting Your AI Tool

This tutorial walks you through adding the Everwall MCP server to the four most common tools: Claude (web and desktop), ChatGPT, Claude Code, and OpenAI Codex. Pick the section for your tool. You only need to do this once.

The connection URL is the same everywhere:

https://mcp.everwall.com/mcp

Every tool uses the same one-time sign-in. You add the server, and the tool prompts you to log in to Everwall through a secure window — there are no keys to copy or paste. Claude, ChatGPT, Claude Code, and OpenAI Codex all work this way.

Note: Menu labels in Claude and ChatGPT shift as those products evolve. If a button name here doesn’t match exactly, look for the nearest equivalent — the overall flow (add a custom connector → paste the URL → connect and approve) stays the same.

Claude (web and desktop)

Custom connectors in Claude are shared across claude.ai, Claude Desktop, and mobile — set it up once and it’s available everywhere you use Claude. Claude connects to Everwall from Anthropic’s cloud, so nothing runs on your machine.

Availability: Free, Pro, Max, Team, and Enterprise. (Free accounts can have only one custom connector.)

Step 1 — Open your connectors. Go to Settings > Connectors (you may reach it via Customize > Connectors).

Step 2 — Add a custom connector. Click the “+” button, then “Add custom connector.”

Step 3 — Enter the details. Give it a Name (for example, “Everwall”) and paste the server URL:

https://mcp.everwall.com/mcp

Leave the Advanced settings (OAuth Client ID/Secret) blank — the Everwall server registers your tool automatically. Click “Add.”

Step 4 — Connect. Back in the connectors list, click “Connect” on your new Everwall connector. An Everwall sign-in window opens. Log in, choose the account(s) and permissions to grant, and approve.

Step 5 — Use it in a chat. In any conversation, open the “+” (attachments) menu, find Connectors, and enable Everwall for that chat. Now you can ask things like “List my walls” or “Moderate the queue on my event wall.”

Team / Enterprise note: an Owner adds the connector for the organization first (Organization settings > Connectors > Add > Custom > Web), then each member connects their own Everwall login from Customize > Connectors.

ChatGPT

Custom MCP servers in ChatGPT are added through Developer Mode, currently in beta. They work in both normal chat and deep research once Developer Mode is on.

Availability: Plus, Pro, Business, Enterprise, and Edu, on ChatGPT web. On Business/Enterprise/Edu, a workspace admin must enable Developer Mode first.

Step 1 — Turn on Developer Mode. Go to Settings > Apps > Advanced settings and enable Developer mode. (On a workspace, an admin enables it under Permissions & Roles and chooses who can create custom apps.)

Step 2 — Create the app. Open Settings > Apps and click “Create app” (this button only appears when Developer Mode is on).

Step 3 — Enter the server URL and auth. Paste:

https://mcp.everwall.com/mcp

Choose OAuth as the authentication type. Everwall supports the automatic registration ChatGPT uses, so there’s nothing else to fill in.

Step 4 — Sign in. ChatGPT opens an Everwall login/consent window. Log in, grant the account(s) and permissions, and approve.

Step 5 — Use it. Everwall’s tools are now available to ChatGPT. Write actions ask for your confirmation before running by default.

Claude Code

Claude Code (the CLI) connects to remote MCP servers over HTTP and signs you in with the same Everwall login flow.

Step 1 — Add the server. This should be done in your terminal, not in Claude Code. If you do it inside claude code, it’ll spawn a bash to run the command and you’ll still have to exit Claude Code and go back into it to use it, so it’s better to just run it directly yourself in the terminal.

claude mcp add --transport http everwall https://mcp.everwall.com/mcp

Step 2 — Log in. Start Claude Code and run: /mcp

The /mcp panel lists the Everwall server and opens a browser to complete the Everwall login and consent. Approve, and you’re connected.

Step 3 — Choose where it’s saved (optional). Use --scope to control who sees the server:

• local (default) — just you, just this project.
• project — shared with your team via a checked-in .mcp.json.
• user — available to you across all your projects.

Example: shared with the team, OAuth

claude mcp add --transport http --scope project everwall https://mcp.everwall.com/mcp

Step 4 — Verify. Run claude mcp list to confirm Everwall is connected. Use claude mcp get everwall for details or claude mcp remove everwall to remove it.

Note: If you’re pasting a config block instead, Claude Code accepts "type": "streamable-http" as an alias for http. An entry with a url but no type is treated as an error, so always include the transport type.

OpenAI Codex

Codex (the CLI) supports remote Streamable HTTP MCP servers directly, no bridge or helper package needed. You add the server in a TOML file, then log in.

Step 1 — Open your config. Edit ~/.codex/config.toml (global) or .codex/config.toml inside a trusted project.

Step 2 — Add the Everwall server.

[mcp_servers.everwall]
url = "https://mcp.everwall.com/mcp"

Step 3 — Log in. Run the Everwall OAuth login:

codex mcp login everwall

This opens a browser to sign in and approve — the same flow as everywhere else.

Step 4 — (Optional) tune timeouts. Ranged stats and bulk operations can take a little longer; the defaults are usually fine, but you can raise them:

[mcp_servers.everwall]
url = "https://mcp.everwall.com/mcp"
startup_timeout_sec = 10   # default 10
tool_timeout_sec = 60      # default 60

Step 5 — Verify. Start Codex and confirm the Everwall tools are available. Run codex mcp --help if you need the full list of MCP subcommands for your installed version.

What to try first

However you connected, a great first prompt is: “Use Everwall to tell me who I am and which walls I can access.”

This runs the whoami and list-walls tools and confirms everything is wired up. From there, head to Working With Your Walls for practical prompts and workflows.