View as Markdown
MCP

MCP Server Overview

Give your AI assistant direct access to your ad accounts

What is MCP?

The Model Context Protocol (MCP) is an open standard that allows AI assistants to securely connect to external tools and data sources. Instead of copy-pasting data between your AI and ad platforms, MCP gives your AI direct, controlled access.

AI-Native

Built for LLMs to understand and use your ad data naturally

Real-time

Live campaign data, not stale exports or screenshots

Conversational

"Pause all campaigns with ROAS below 2" just works

How it works

1

Configure your AI client

Point Claude Desktop, Claude Code, ChatGPT, Cursor, Codex, Amp, or Windsurf at https://mcp.syntermedia.ai
2

Authenticate with your API key

Pass it as the X-Synter-Key header — your API key gives the AI access to your connected ad accounts
3

Ask naturally

"What's my Google Ads spend this month?" — the AI handles the rest

Quick setup (Claude Code)

Run this in your terminal:

bash
claude mcp add synter-ads \
  --transport http \
  https://mcp.syntermedia.ai \
  --header "X-Synter-Key: syn_your_api_key_here"

Get your API key from the Developer Portal, or run synter_onboarding_start with your email if you don't have an account yet — no sign-up flow required first.

Example prompts

Show me my top 5 campaigns by ROAS this month
What's my total ad spend across all platforms?
Pause all campaigns with CTR below 1%
Increase budget by 20% for campaigns with ROAS above 4

Supported AI clients

Every client below connects to the same https://mcp.syntermedia.ai endpoint with an X-Synter-Key header — see Configuration for the exact snippet per client.

Pro tip

Use the get_connection_status tool (free) to verify your ad accounts are connected before running paid operations.

No keyless tier

Unlike some MCP servers, Synter doesn't offer an anonymous/keyless slice of tools — every tool call needs a real API key. If you don't have an account yet, call synter_onboarding_start with your email to get one without visiting the site first.
Was this page helpful?