What it is and what it is for
If you already use an AI assistant every day (Claude, Claude Code, Codex, Cursor, VS Code or ChatGPT), you can connect Lera to it. From then on you ask for things in one sentence, for example “summarise yesterday’s calls” or “make the agent ask for the postcode before booking”, and the assistant does it in your account.
The connection uses the standard these assistants follow to talk to other platforms, called MCP. You do not need to install anything or copy any key. The connector is included in every Lera plan.
How to connect it in Claude
1. In Claude, open Settings and go to Connectors (in some versions it is under Customize). Press the button to add a custom connector.
2. Give it a name, for example Lera, and paste the address https://leracore.com/mcp.
3. Keep the options Claude detects on its own: sign in through OAuth before using it, and register the client automatically. Do not add any header or key.
4. Press Add and then Connect. A Lera page opens.
5. Sign in as usual, check the project and choose View and manage or View only. Press Allow and you go back to Claude, already connected.
In Claude business accounts, whoever manages the organisation adds it first, and then each person connects it with their own Lera account.
How to connect it in Claude Code
If you sign in to Claude Code with your Claude account and already connected it in Claude, there is nothing else to do: it shows up when you type /mcp.
Otherwise, in a terminal: claude mcp add --transport http --scope user lera https://leracore.com/mcp. Then open Claude Code, type /mcp, pick lera and follow the sign-in in your browser.
How to connect it in Codex
In a terminal: codex mcp add lera --url https://leracore.com/mcp and then codex mcp login lera, which opens the browser to sign in.
In the Codex app or extension: Settings, MCP servers, Add server, Streamable HTTP. Paste the address, save and press Authenticate. The terminal, the app and the extension share the connection.
How to connect it in Cursor and VS Code
In Cursor, add lera to the file ~/.cursor/mcp.json with the address https://leracore.com/mcp?toolset=construir, without headers, and connect it from Settings, MCP. The ?toolset=construir keeps the agent-building tools, because Cursor allows about 40 across all its servers.
In VS Code, add an HTTP server with the address https://leracore.com/mcp (in .vscode/mcp.json or your user one). When it starts, the browser opens to sign in.
The steps for each one, with what to copy, are also in Lera, in Settings, API Keys, AI assistants tab.
How to connect it in ChatGPT
Your own connectors are added from Settings, Apps, in developer mode, with the same address https://leracore.com/mcp and signing in as the way in. Then the same Lera page opens as in Claude.
Which ChatGPT plans allow it is decided by OpenAI and changes often: some only allow viewing. If you do not see the option, your plan does not have it yet.
What it can do and with what permission
With View only it can see your agents, your calls and your analytics, without changing anything. With View and manage it can also create and change agents, functions and knowledge bases, test an agent in writing and publish it when you ask.
It always acts with your role in the project. If you are an Editor, it does not touch phone numbers or outbound calls, just like in the panel. If one day you leave the team, the connection stops working on its own.
You need to be an Administrator or an Editor of the project to connect it. What it spends testing an agent is charged the same as testing it in the panel.
One tip. In your assistant you can choose which tools ask you to confirm before being used. Keep confirmation on the ones that change things, such as publishing an agent, so you see what it will do before it does it.
How to remove its access
From Lera, in Settings, API Keys, AI assistants tab, under Connected apps. There you see every assistant with access, since when and when it was last used, and you revoke it with one click. The administrator sees the whole team’s; each Editor, their own.
You can also disconnect it from the assistant itself. API keys keep working the same and are for programs with nobody in front of them: your ERP, n8n, your automations or a server without a browser.