irinabuht12-oss/google-ads-meta-ads-mcp

Google Ads MCP server + Meta Ads MCP (Facebook Ads MCP) + Google Analytics MCP (GA4) + Search Console in one hosted remote MCP for Claude, ChatGPT, Cursor & n8n: OAuth login, no API keys, approval-gated writes, 7-day free trial. By Ryze AI.

What is this repository?

google-ads-meta-ads-mcp is the source code for a remote Model Context Protocol (MCP) server that exposes three major advertising and analytics APIs as “tools” that can be called by AI assistants such as Claude, ChatGPT, Cursor, n8n, Windsurf, etc.

Google Ads API (campaign, keyword, bidding, reporting, etc.)
Meta (Facebook/Instagram) Marketing API (campaign, ad‑set, creative, audience, reporting, etc.)
Google Analytics 4 Data API (standard, realtime, audience, property configuration, etc.)

The server implements the open‑source Model Context Protocol so that an LLM can invoke these tools via natural‑language prompts. Ryze AI hosts the server at https://connector.get-ryze.ai/mcp; the repo contains the code, documentation, and a catalog of the 250+ available tools.

Who might use it?

Role	Why it matters
Marketers / Agencies	Manage Google Ads, Meta Ads, and GA4 from a single chat‑based interface, with write‑access that is gated by an approval step.
LLM‑powered workflow builders	Plug the MCP endpoint into Claude, ChatGPT, Cursor, n8n, Windsurf, etc., to automate ad creation, budget adjustments, performance audits, and cross‑platform reporting.
Developers of AI agents	Use the MCP spec to add a rich advertising toolbox to custom agents without handling OAuth, API keys, or separate SDKs.
Product teams	Offer a hosted “ads‑as‑a‑service” capability to customers who want AI‑driven campaign management without self‑hosting the APIs.

Core features (as listed in the README)

Google Ads (≈150 tools)

Campaign CRUD for Search, Display, Shopping, Performance‑Max, Demand‑Gen, Video, App
Keyword research via Keyword Planner (volume, CPC, competition)
Bidding & budget tools, portfolio strategies, bid adjustments
Creative creation (responsive search/display ads, demand‑gen ads, app ads)
Audience targeting (custom, remarketing, customer‑match, combined)
Full performance analytics (metrics, ROAS, search‑terms, negative‑keyword suggestions)
Extensions (sitelinks, callouts, calls, images, prices, promotions, lead forms)
Experiments & A/B testing
Conversion‑tracking setup and upload

Meta Ads (≈80 tools)

Campaign, ad‑set, and creative CRUD across Facebook & Instagram
Budget, schedule, placement, and optimization controls
Audience tools (custom, look‑alike, interest/behavior/demographic)
Lead‑generation forms and product‑catalog management
Page management (posts, boosted posts, insights)
Business‑account administration (users, pages, pixels)
Detailed performance insights with breakdowns

Google Analytics 4 (≈20 tools)

Standard, pivot, and realtime reports
Audience, segment, and channel‑group listings
Property configuration (data streams, enhanced measurement, retention)
Attribution model settings
Event & conversion management
Integration links (BigQuery, DV360, Search Ads, Firebase)
Property‑level annotations and metadata queries

Cross‑platform capabilities

Side‑by‑side ROAS comparison between Google and Meta campaigns
Correlation of ad spend with GA4 conversion events
Unified budget‑allocation recommendations

How to connect (quick‑start)

Add the connector URL (https://connector.get-ryze.ai/mcp) in the AI assistant’s Connectors UI (Claude Desktop, Claude Code, ChatGPT, Cursor, Windsurf, n8n, etc.).
Sign in with the Google account that owns the Google Ads/GA4 properties or the Facebook account that owns the Meta ad accounts. The server handles OAuth 2.1 + PKCE; no developer token or API key is required.
Approve writes – any operation that changes a campaign, budget, or creative is staged and must be approved through the Ryze UI before it is sent to the underlying APIs.
Start prompting – ask the assistant to list campaigns, create a new ad, pull a GA4 report, etc. The assistant translates the natural‑language request into the appropriate MCP tool call.

The README provides per‑assistant setup guides (docs/SETUP_*.md) and example prompts for each platform.

Limitations & considerations

Hosted service – the server runs on Ryze’s infrastructure; you must use their subscription (Paid Ads plan $89 / month, 7‑day free trial). The code is open‑source, but the hosted endpoint is a paid product.
Write gating – all write operations go through an approval step to prevent accidental budget changes. This adds a manual checkpoint.
OAuth scope – the connector can only act on accounts the logged‑in user has permission to manage.
Rate limits – underlying Google and Meta APIs enforce their own quotas; heavy automation may require additional quota requests.
Feature parity – while the list is extensive, some niche Google Ads or Meta Marketing API features may not yet be wrapped as MCP tools.

TL;DR

google-ads-meta-ads-mcp bundles Google Ads, Meta Ads, and GA4 APIs into a single remote MCP server that AI assistants can call as natural‑language tools. It removes the need for separate API keys, developer tokens, or self‑hosted servers, offering read/write access (with approval) for marketers and AI‑agent developers. The repository contains the server implementation, a catalog of 250+ tools, and step‑by‑step connection guides for popular LLM interfaces.