How to connect
The Moonlit MCP server works in every AI app that supports remote MCP servers (in MCP terms, a client). Add the server URL, sign in with your Moonlit account, done.
https://mcp.moonlit.ai/mcpMost clients use the same flow: the client registers with the server, opens the Moonlit sign-in page in your browser, and you approve the connection with your Moonlit account. No key is needed. If you do not have an account yet, create one at app.moonlit.ai. The exceptions authenticate with an organization MCP key instead: Microsoft 365 Copilot, deployed services, and pooled organization usage. See Organizations.
Claude
Docs- In Claude on the web, go to Settings, then Connectors.
- Click Add custom connector, name it Moonlit, and paste the server URL.
- Click Add, sign in with your Moonlit account, and approve the connection.
Connectors sync across Claude on the web, the desktop app, and the mobile apps on the same account. On Claude Team and Enterprise plans, custom connectors are added by a workspace Owner under organization settings; members then sign in individually.
Microsoft 365 Copilot
DocsMicrosoft 365 Copilot connects through an agent built in Microsoft Copilot Studio.
- In Microsoft Copilot Studio, open your agent (or create one).
- Go to Tools, select Add a tool, then New tool, and choose Model Context Protocol.
- Name the server Moonlit, paste the server URL, and authenticate with an organization MCP key (see Organizations).
- Add the Moonlit tools to the agent and publish it. Published to the Microsoft 365 Copilot channel, the agent (and Moonlit with it) is available to your colleagues inside Copilot chat.
ChatGPT
Docs- ChatGPT supports remote MCP servers through Developer mode, available on paid plans in the web app.
- Enable Developer mode in Settings, under the apps or security section of your account.
- Add a new connector with the server URL, choose OAuth authentication, and sign in with your Moonlit account.
Langdock
Docs- In Langdock, go to Integrations, click Add Integration, then Start from scratch.
- Select Connect remote MCP and paste the server URL.
- Choose OAuth as the authentication method, click + Add connection, and sign in with your Moonlit account.
- Click Test connection, select the Moonlit tools, and click Save tools.
Langdock converts the discovered tools into actions, which you can then enable in chats and assistants. Workspace admins can share the integration with the whole workspace; each member signs in with their own Moonlit account.
Le Chat (Mistral)
DocsCustom connectors in Le Chat are managed by a workspace administrator. On individual plans, the account owner is the administrator by default.
- Open the Connectors page, click + Add Connector, and switch to the Custom MCP Connector tab.
- Name the connector moonlit (connector names cannot contain spaces) and paste the server URL.
- Le Chat detects the authentication method automatically. Complete the sign-in with your Moonlit account when prompted.
Perplexity
DocsCustom remote connectors are available on paid Perplexity plans (Pro, Max, and Enterprise).
- Click your profile icon, then Settings, then Connectors.
- Click + Custom connector and select Remote.
- Name it Moonlit, paste the server URL, choose OAuth authentication, and acknowledge the third-party connector notice.
- Sign in with your Moonlit account to finish.
Claude Code
Docsclaude mcp add --transport http moonlit https://mcp.moonlit.ai/mcpThen run /mcp inside a session to trigger the sign-in, and verify with claude mcp list. Add -s user to make the server available in every project, or -s project to share it with your team via a checked-in .mcp.json.
Cursor
DocsAdd the server to ~/.cursor/mcp.json (all projects) or .cursor/mcp.json (one project):
{
"mcpServers": {
"moonlit": {
"url": "https://mcp.moonlit.ai/mcp"
}
}
}Open Settings, then MCP, and click the authentication prompt next to the Moonlit entry to sign in.
VS Code (GitHub Copilot)
DocsAdd the server to .vscode/mcp.json in your workspace, or run MCP: Open User Configuration from the command palette for a global setup. Note the root key is servers:
{
"servers": {
"moonlit": {
"type": "http",
"url": "https://mcp.moonlit.ai/mcp"
}
}
}VS Code starts the sign-in flow automatically the first time Copilot uses the server in agent mode. On Copilot Business and Enterprise, an organization admin must first enable the MCP policy; it is off by default.
Codex
Docscodex mcp add moonlit --url https://mcp.moonlit.ai/mcp
codex mcp login moonlitOr add it to ~/.codex/config.toml directly:
[mcp_servers.moonlit]
url = "https://mcp.moonlit.ai/mcp"The Codex IDE extension shares the same configuration file.
Gemini CLI
Docsgemini mcp add --transport http moonlit https://mcp.moonlit.ai/mcpOr add it to ~/.gemini/settings.json. Note that remote servers use the httpUrl key:
{
"mcpServers": {
"moonlit": {
"httpUrl": "https://mcp.moonlit.ai/mcp"
}
}
}Sign in with /mcp auth moonlit inside a session. Gemini Code Assist in VS Code reads the same settings file.
Windsurf
DocsAdd the server to ~/.codeium/windsurf/mcp_config.json. Note that Windsurf uses the serverUrl key:
{
"mcpServers": {
"moonlit": {
"serverUrl": "https://mcp.moonlit.ai/mcp"
}
}
}
Zed
DocsAdd the server to your Zed settings.json under context_servers:
{
"context_servers": {
"moonlit": {
"url": "https://mcp.moonlit.ai/mcp"
}
}
}Zed prompts for the sign-in flow when no credentials are configured.
Other clients
Any client that supports remote MCP servers over streamable HTTP with OAuth sign-in works. The common configuration shape is:
{
"mcpServers": {
"moonlit": {
"url": "https://mcp.moonlit.ai/mcp"
}
}
}The exact key names vary by client (url, httpUrl, serverUrl); check your client's MCP documentation. Clients that cannot complete a browser sign-in (headless or deployed setups) authenticate with an organization MCP key instead; see Organizations.
If something fails
- Verify the URL ends in
/mcp. - Remove and re-add the server entry, then complete the sign-in again.
- Check your plan and remaining quota by asking your assistant to call
get_account_status. - Still stuck? Email support@moonlit.ai with the client you are using and what you see.