Create a workspace API key
In console Settings → Workspace → API keys, create a key and copy the MCP config snippet once.
Remote MCP
Inspect traffic in your IDE.
Connect an MCP client to Mercur’s remote MCP server and let an agent list tunnels, search HTTP request history, and open full headers and bodies — without leaving your editor.
Webhook debugging stops at the browser tab
You already have request history in Mercur. When an AI agent is helping you fix a failed delivery, it should be able to list recent requests and open the payload that actually arrived — not ask you to paste screenshots.
SetupStep 1 Step 2 Step 3
From key to agent-readable history
Add it to your MCP client
Paste the snippet as a remote / HTTP MCP server pointing at console …/api/mcp with Authorization: Bearer …
Ask the agent to inspect
list_servers → list_requests → get_request. Summaries omit bodies; detail includes headers and bodies.
Full walkthrough: Debug webhooks with Cursor MCP.
Tools
What the agent can call
Three read-only tools. Summaries stay light; full payloads only when you ask for one request.
- list_servers() → { items }
- Lists HTTP local tunnels in the key’s workspace.
- list_requests(serverId, filters?) → { items, nextCursor }
- Paginated request summaries (method, path, status, timing) with filters — no bodies.
- get_request(serverId, requestId) → { item }
- Full request/response detail including headers and bodies for one log id.
FAQ
Remote MCP questions
What is Mercur remote MCP?
A Streamable HTTP MCP server on the Mercur console that exposes read-only tools for tunnel request history. Any MCP client that supports remote HTTP servers (Cursor, Claude Desktop, and others) can connect with a Bearer workspace API key — no local npm package.
Which plans include MCP?
Solo and Team. On Free, Settings shows an upgrade path to Pricing before you can create a workspace API key.
How do I connect an MCP client?
Create a workspace API key in Settings, copy the generated mcp.json snippet (url + Authorization header), and paste it into your client as a remote / HTTP MCP server. See the debug-webhooks-with-cursor-mcp guide for the full walkthrough.
Can a key read another workspace?
No. Keys are workspace-scoped. list_servers and request tools only see tunnels belonging to that workspace.
Let your agent see the real webhook
Create a workspace API key once. Your agent lists tunnels, finds the failed delivery, and reads the payload — without leaving the editor.