Terminal & automation

Use MCP Ads from the terminal

Two ways in, depending on what you are building. Add the remote server to a terminal MCP client and talk to it in plain language, or call the REST API directly from a script when you want deterministic automation rather than a conversation.

There is nothing to install

MCP Ads is a remote server. Your client connects to it over HTTPS and authorises once in the browser, so there is no package to install, no process to keep running, and no key to manage for the conversational path.

Note

Earlier versions of this page described a local stdio CLI installed with npx. That package is not published, so those instructions could not work. The remote server below is the supported path.

Option 1 — a terminal MCP client

Best when you want to ask questions and have the client choose the tools. Claude Code takes one command; editor clients take a small config entry.

Run once in any terminal. Saved to ~/.claude.json and persists across sessions.

Terminal

claude mcp add --transport http mcp-ads https://mcp-ads.com/mcp-slim

Per-client detail, including what to check when tools do not appear, lives in the documentation.

Option 2 — the REST API

Best when you want a script to do the same thing every time. This path uses an API key rather than the browser OAuth flow, and calls the same tools the conversational path uses.

Connect your ad accounts
The API reaches whatever your account has connected. If nothing is connected, every call returns an empty result set.
Create an API key
From your dashboard. The key is shown once and never again, so store it before closing the dialog. Keys can be revoked at any time without affecting your other connections.
List what is available
The manifest returns the tool categories your key can reach.
Terminal
```
curl https://mcp-ads.com/api/cli/manifest \
  -H "Authorization: Bearer mcp_live_YOUR_KEY"
```

Find an action

Each category holds many actions. Search within one rather than listing everything.

Terminal

curl -X POST https://mcp-ads.com/api/cli/list_actions \
  -H "Authorization: Bearer mcp_live_YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{"category": "google_ads_action", "search": "campaign", "limit": 20}'

Call it

Dispatch takes the category, the action name, and that action’s parameters.

Terminal

curl -X POST https://mcp-ads.com/api/cli/dispatch \
  -H "Authorization: Bearer mcp_live_YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{"category": "google_ads_action", "action": "gads_account_overview", "params": {}}'

Careful

An API key is a bearer credential: anything holding it can act on your connected ad accounts. Keep it out of source control and out of client-side code, and revoke it the moment it leaks.

Which one should you use

	MCP client	REST API
Auth	Browser OAuth, once	API key you create and store
Picks the tools	The AI client does	You do, by name
Best for	Analysis, audits, anything exploratory	Scheduled jobs and repeatable automation
Setup	One URL or one command	A key plus your own HTTP calls

Where to go next

Quickstart

Connect an account and run a first query.

Claude Code setup

The terminal client in detail.

Tool reference

Every action the API can dispatch.

Ready to connect?

Connect the platforms you use, then drive them from the terminal or from a script.

Open dashboard Read the guides ?