Docs · Build with agents

Install in your MCP client

Exact steps for Claude, ChatGPT, Cursor and any other MCP client.

Reading as an agent? Every article is also served as Markdown — the full corpus is llms.txt, versioned with these docs.

The A1 Lane MCP server is one URL. Each client below adds it in its own way; the sign-in afterwards is the same everywhere.

urlServer URL

https://a1lane.ai/mcp/v1

Claude

Claude calls a remote MCP server a custom connector.

Open Customize → Connectors and choose Add custom connector.
Name it A1 Lane and paste https://a1lane.ai/mcp/v1 as the server URL.
Leave the sign-in settings as Claude detects them: Sign in now, with Use Claude's published identity (our server supports it; Register automatically works too). Choose Add.
Choose Connect, sign in to A1 Lane, pick the workspace and the scopes.
In a chat, open + → Connectors and switch A1 Lane on.

On a Team or Enterprise plan an owner adds it once under Organization settings → Connectors → Add → Custom (type: Web); members then find it under Customize → Connectors and choose Connect.

Steps follow Claude's custom connector guide as of 2 Nov 2026.

ChatGPT

ChatGPT adds remote MCP servers as plugins, on the web. Your workspace's settings decide whether custom MCP servers are allowed.

Open Plugins (chatgpt.com/plugins), choose +, then Add custom MCP server.
Name it A1 Lane. Under Connection, enter https://a1lane.ai/mcp/v1 as the Server URL.
Set authentication to OAuth. Read the risk warning, choose I understand and want to continue, then Create as a plugin.
Install the plugin and sign in to A1 Lane when asked: pick the workspace and the scopes.
In a new chat, type @ and pick A1 Lane.

Steps follow OpenAI's custom MCP server guide as of 2 Nov 2026.

Cursor

One click, no review needed: Cursor installs the server from a link.

Add to Cursor

The link opens Cursor and asks you to confirm the install. Cursor then opens the A1 Lane sign-in page.

Or add it by hand to ~/.cursor/mcp.json (every project) or .cursor/mcp.json (one project):

jsonmcp.json

{
  "mcpServers": {
    "a1lane": {
      "url": "https://a1lane.ai/mcp/v1"
    }
  }
}

Cursor signs in with OAuth (authorization code with PKCE) and registers itself with our server, so there is no client id to enter.

Steps follow Cursor's MCP docs as of 2 Nov 2026.

Any other MCP client

Anything that speaks remote MCP can connect. What it needs to know:

Server URL: https://a1lane.ai/mcp/v1
Transport: Streamable HTTP (JSON-RPC over POST)
Sign-in: OAuth 2.1 with PKCE (S256). An unauthenticated request gets a 401 that points to /.well-known/oauth-protected-resource; the authorization server's metadata is at /.well-known/oauth-authorization-server.
Client registration: Client ID Metadata Documents or Dynamic Client Registration — no pre-registration needed.
Scopes: read, write, publish

Steps follow the MCP authorization spec as of 2 Nov 2026.

Signing in, or using a token

Every client above signs in through the normal A1 Lane sign-in page: you sign in (or sign up), pick one workspace and choose what the assistant may do — read, write, publish. Drafts stay the default; nothing publishes without the publish scope and a schedule.

If your client can't do OAuth, a workspace token works too. Create one in Settings → API & Webhooks → Create a token, then send it as a header on every request:

httpHeader

Authorization: Bearer plt_your_workspace_token

In Cursor, add that header under headers in mcp.json. Treat the token like a password: anyone holding it can act in that workspace until you revoke it.

Check it worked

Ask: “Which A1 Lane workspace am I in, and what's next on my calendar?” The assistant calls list_accounts for the workspace, then list_posts with status: "scheduled" for what's next (unfiltered, list_posts lists newest first across every status). Nothing is written.

If it names the wrong workspace, disconnect, reconnect and pick the right one on the sign-in page. If it sees no channels, connect one in the web app first.