Remote MCP
Remote MCP lets an AI agent in your editor inspect HTTP request history on your Mercur tunnels. You connect once with a workspace API key — no local install and no npm package.
Use it when you are debugging webhooks or API traffic and want the agent to list recent requests, open a failed delivery, and read the headers and body that actually arrived.
1. Create a workspace API key
Sign in to the console, open Settings → Workspace → API keys, and create a key. Any workspace member can create or revoke keys.
Copy the secret immediately — Mercur shows the full token once. The page also gives you a ready-made config snippet for remote MCP clients.
Open Workspace Settings
2. Add Mercur to your MCP client
Paste the snippet into your client as a remote / HTTP MCP server. The shape looks like this:
{
"mcpServers": {
"mercur": {
"url": "https://console.mercur.sh/api/mcp",
"headers": {
"Authorization": "Bearer mcr_…"
}
}
}
}
Replace the token with yours. On localhost, use your console origin instead of
https://console.mercur.sh.
Works with Cursor, Claude Desktop, and other clients that support remote MCP over HTTP.
3. Ask the agent to inspect traffic
Once the server is connected, ask the agent to list your tunnels and inspect recent requests — for example a failed webhook or an unexpected status code.
The agent can only see tunnels and logs in the workspace that owns the API key.
Tools
Mercur exposes three read-only tools:
list_servers— list HTTP local tunnels in the workspacelist_requests— browse request history as summaries (method, path, status, timing). No bodies or headers in the list. Optional filters: path, method, status, session, time range, and paginationget_request— open one request in full, including headers and bodies
Typical flow: list_servers → list_requests → get_request for the delivery you care
about.
Revoking access
Revoke a key anytime from Settings → Workspace → API keys. The client stops working with that token immediately. Create a new key if you need to reconnect.
Related
- Product overview: /mcp
- Step-by-step guide: /guides/debug-webhooks-with-cursor-mcp
- Plans and pricing: /pricing