Call Meltwater MCP from a shell

Call the public MCP from a terminal when you need a script or a local bridge. Every client connects to the same Meltwater MCP server, so the URL and sign-in are the same wherever you use it.

Connection details

Setting	Value
Server URL	https://api.meltwater.com/v2/mcp
Transport	Streamable HTTP
Authentication	OAuth sign-in at app.meltwater.com, or a Meltwater API token in the apikey header

Before you start

• A Meltwater account with access to Meltwater MCP.
• The server URL, exactly https://api.meltwater.com/v2/mcp, with no trailing slash.
• A Meltwater API token for the apikey header, or a client that can complete the browser sign-in.
• curl for a direct call, or Node.js 18 or later with npx if the client only speaks stdio.

Connect

1. Call the server directly

You don't need to install a Meltwater CLI or package. POST an MCP initialize request to the public URL. The Building agents guide sends the API token in an apikey header. A JSON-RPC result means the endpoint accepted the token. A request with no credential returns 401 and the body no bearer token. The example uses MCP protocol version 2025-03-26.

Terminal

curl -sS https://api.meltwater.com/v2/mcp \
  -H "Content-Type: application/json" \
  -H "Accept: application/json, text/event-stream" \
  -H "apikey: <YOUR MELTWATER API KEY>" \
  -d '{
    "jsonrpc": "2.0",
    "id": 1,
    "method": "initialize",
    "params": {
      "protocolVersion": "2025-03-26",
      "capabilities": {},
      "clientInfo": { "name": "meltwater-manual", "version": "1.0.0" }
    }
  }'

2. Bridge a stdio-only tool

If a local agent only launches a command, point mcp-remote at the public URL. The config pins mcp-remote@0.14.3 so npx does not install a later release. This JSON does not attach a token. The public server still returns 401 until authentication succeeds. Prefer a remote HTTP config in any client that supports one.

stdio bridge

{
  "mcpServers": {
    "meltwater": {
      "command": "npx",
      "args": ["-y", "mcp-remote@0.14.3", "https://api.meltwater.com/v2/mcp"]
    }
  }
}

3. Confirm the tool list

Open a new conversation and ask the assistant to list the Meltwater tools it can call.

An empty list means either the account can't see any tools, or the client is still using a failed connection. After you fix authentication, start a new chat so the client runs initialize again.

Check the connection

Once the server shows as connected, paste this prompt into a new conversation. It only uses tools the client has actually listed.

First session

I just connected Meltwater. Use the Meltwater MCP tools if they are available.

1. Confirm the connection by listing the tools you can call. If none are listed, say so and stop.
2. Ask me which brand, company, or topic to brief if I have not named one.
3. Use the available tools to pull recent coverage and the main narratives. Cite the source and date for each material fact the tools return.
4. Return a short briefing: what is being said, what changed, what is unresolved, and one recommended next action.
5. Before calling a tool, check its schema. Ask only for missing inputs, in a single message, then name the tools you will call.

See which tools you can access

Use the Tools Explorer to browse the tools available to your account - each tool's name, description and parameters - without connecting a client.

If it does not connect

The URL is rejected

Use HTTPS and the exact URL https://api.meltwater.com/v2/mcp. A trailing slash, an http:// scheme, or any other Meltwater URL won't reach the server.

The client only offers SSE

Choose Streamable HTTP, which uses POST. An unauthenticated GET, POST or DELETE to the server URL returns 401 with the body no bearer token.

Tools stay empty after sign-in

Start a new chat so the client runs initialize and tools/list again. Check that the Meltwater account you signed in with has access to Meltwater MCP.

Still stuck?

Take a look at the FAQs, or contact our Support team.