Last updated May 2026
Meet the Google Ads MCP alternative that does more than read data
Google Ads MCP is a free, open-source server that lets your AI agent query Google Ads data via GAQL. Good for pulling reports, but read-only by default, Google-only, and requires Python + OAuth setup. AdKit covers Google, Meta, and TikTok with full read+write MCP, a draft-first safety system, a 500,000+ ad library, competitor tracking, and AI creative tools. No GAQL knowledge needed, ready in two minutes.
The short version
| Dimension |   | |
|---|---|---|
| What it is | A complete Ads toolbox with a built-in MCP. | A free, open-source MCP server for querying Google Ads data. |
| MCP Capabilities | Launch and manage campaigns, track competitors & their ads, browse 500,000+ ads, generate static ads. | Read Google Ads data via GAQL queries. List accounts. Get resource metadata. Read-only by default. |
| Platform coverage | Meta, Google, and TikTok (MCP). Meta, Google, LinkedIn (ad library). | Google Ads only. No other platforms. |
| Can your agent create campaigns? | Yes. Draft-first: agents write inside AdKit, you review and publish with one click. | No (canonical repo). Experimental mutation tools exist in a separate repo, disabled by default, labeled 'not for production use'. |
| Who is it for? | Marketers, founders & agencies who want to give advertising capabilities to their AI agents. | Developers with Google Cloud experience who need to query Google Ads data from an LLM. |
| Starting price | AdKit starts at $29/month for 1 account, MCP included. $49/month for unlimited accounts. (Free trial available) | Free and open source (Apache 2.0). Self-hosted. Requires own Google Cloud project and developer token. |
or ask an AI to help you choose:
"Other tools I found were just proxies for the ads API. One bad call and you could mess up a live campaign. What I like about AdKit is that everything goes through a draft first. That buffer between your agent and your live account is what makes it safe to use."
Nevo David
Founder of Postiz
The complete comparison
Read-only vs. full workflow
Google Ads MCP does one thing: let your AI agent read Google Ads data. It executes raw GAQL queries and returns results. No campaign creation, no bid adjustments, no creative uploads. A separate experimental repo adds mutation tools, but Google labels it "not intended for production use" and disables them by default.
AdKit gives your agent the full loop. Read performance data, create campaigns, adjust targeting, upload creative, manage budgets. All across Meta, Google, and TikTok through task-shaped MCP tools your agent can use without knowing platform API internals.
If you only need to pull reports, Google's MCP works. If you need to act on what you find, AdKit handles the execution layer too.
Google-only vs. cross-platform
Google Ads MCP connects to one platform: Google Ads. If you also run Meta or TikTok campaigns, you need separate MCP servers for each, with separate auth flows and separate maintenance.
AdKit covers Meta, Google, and TikTok in a single MCP connection. One setup, one tool surface, one place your agent goes to manage all your campaigns. The ad library also tracks competitors across Meta, Google, and LinkedIn, so research and execution live together.
If you only run Google Ads and only need to read data, the Google MCP is fine. If you run ads across platforms and want to act on what you find, you need something broader.
Setup: 2 minutes vs. 60 minutes
Google Ads MCP requires Python 3.10+, a Google Cloud project, an OAuth 2.0 client, a developer token, and a credential file. The developer token requires a separate application via Google's API Center, with review taking days to weeks. After that, local setup takes 15-20 minutes for experienced developers. Cloud Run deployment took one developer 2 extra days of troubleshooting.
AdKit: sign up, connect your ad account via OAuth, done. Your MCP is ready in your AI client within two minutes. No Python, no credential files, no Cloud projects, no application process.
If you have the technical chops and patience for Google's approval process, their setup is manageable. If you'd rather spend those days running ads, AdKit gets you there in two minutes.
Where Google Ads MCP wins
It's free. Open source, Apache 2.0, no subscription. If budget is the only factor, hard to beat zero.
It's first-party Google. Built by the Google Ads API team, so you get the deepest possible data access with zero ban risk. Raw GAQL gives you custom cross-account reporting, automated anomaly detection, and full access to every metric and segment Google exposes.
If you're a developer who only needs Google Ads data access and you're comfortable with GAQL and self-hosting, this is a solid, free tool. AdKit targets a different job.
What happens if you enable mutations?
The canonical Google Ads MCP is read-only, so safety is simple: your agent can't break anything because it can't change anything.
But if you need write access and enable mutations through the experimental repo, there are no guardrails. No draft system, no confirmation gates, no undo. Tools execute directly against your live Google Ads account the moment your agent calls them.
AdKit uses draft-first architecture. Every change lands as a draft. You review in the dashboard, approve, one click publishes. Your live account stays untouched until you say go.
"If you've ever tried any other MCP, you'll quickly realize that they might give you access to your data, but the moment you start creating campaigns, everything will break. Adkit created something that just works like magic."
Flo Pariset
Founder of Notis AI
Still struggling to choose?
Pick AdKit if you want to do more than read data.
You want to create, manage, and optimize campaigns through your AI agent, not just query metrics.
You run ads on Meta, Google, or TikTok and want one MCP for all of them.
You want the full loop: competitor tracking, ad library, creative tools, campaign launch.
You want draft-first safety so your agent can't touch your live account without approval.
That's marketers, founders running their own ads, growth leads, agencies.
Pick Google Ads MCP if you just need free Google Ads data access.
You only run Google Ads and only need to read performance data.
You're a developer comfortable with Python, OAuth, GAQL, and self-hosting.
You want free and open-source tooling you can fork and extend.
You need first-party Google data access with zero third-party trust required.
If that's you, Google Ads MCP is a well-built, free tool for the job it covers.
"Now I can use my AI agent for monitoring competitors, creating static ads, publishing them, and analyzing their performance. This has 10x'd my capabilities to iterate on ads."
Flo Pariset
Founder of Notis AI
A word from the founder
Hi there 👋 I'm Nico, the founder of AdKit 😄
Before building startups, I was a media buyer. I managed over $1,000,000 in Ads, sold two startups grown with ads, and helped 1,000+ makers learn Meta Ads through my online guides.
I love ads but... it involves a lot of repetitive, boring work 😩 And I hated every second of it. So I built AdKit to fix that.
To let me and other marketers focus on the fun parts: the strategy, the thinking, the creativity. And not have to waste my time clicking around in buggy interfaces.
It's the app I always dreamed of, and I hope you'll like it too 😀
Everything side-by-side
Product category & scope
| Feature | | |
|---|---|---|
| Product type | Full ads toolbox | Open-source data layer |
| Primary job | Competitor tracking + creative + launch in one loop | Query Google Ads data via GAQL from an LLM |
| Target audience | Marketers, growth leads, agencies | Developers, data analysts, Google Ads API users |
Company & trust
| Company location | Singapore | United States |
| Data hosted in | Europe | SELF-HOSTED (USER'S INFRA) |
| Founder background | Nico Jeannen, ex media buyer. Managed $1M+ in ad spend, sold two startups grown with ads, taught 1,000+ makers Meta Ads. | Google Ads API Team. Author: Dean Lukies. |
| Trains AI on your data | No. | No (self-hosted, data stays local) |
| Date founded | April 2025 | October 2025 |
| Funding | Independent & 100% Bootstrapped | Google (Alphabet) |
| Open Source | No | Yes |
Agent safety & architecture
| Approved by ads platforms | Yes | Built by Google themselves |
| Safety model | Agent creates drafts you can review in the dashboard, then publish with one click | Read-only by default. Experimental mutations execute directly on live account with no confirmation. |
| Draft system for all operations | Yes | No |
| Agent can touch live account | Only with user approval | Cannot (read-only canonical). Yes without confirmation (experimental mutations). |
| Campaign creation | Yes | Experimental, disabled by default |
MCP & platform coverage
| MCP architecture | Built for AI agents: merged tools, sensible defaults, with raw platform APIs available optionally | 3 tools + resources. Raw GAQL execution. User must know Google Ads Query Language. |
| MCP tools | Task-shaped tools across Meta, Google, and TikTok. Sensible defaults, raw API passthrough available. | 3 tools (search, list_accessible_customers, get_resource_metadata). Raw GAQL queries. |
| Self-hosting required | No (hosted SaaS) | Yes (local or Cloud Run) |
| Meta Ads | Yes | No |
| Google Ads | Yes | Yes |
| TikTok Ads | Yes | No |
| Soon | No | |
| Reddit Ads | Soon | No |
| X Ads | Soon | No |
Supported agents & clients
 Claude | Yes | Yes |
 Claude Code | Yes | Yes |
 Codex | Yes | Unknown |
 Cursor | Yes | Yes |
 ChatGPT | Yes | Unknown |
 Gemini CLI | Yes | Yes |
 Perplexity | Yes | Unknown |
 OpenClaw | Yes | Unknown |
 Hermes | Yes | Unknown |
 n8n | Yes | No |
Competitor research
| Drop-in competitor tracking by URL | Yes, auto-scrapes every ad, continuously | No |
| Tracked ad library size | 500,000+ ads across Meta, Google, LinkedIn | No |
| New-ad alerts | Yes | No |
| Swipe file / saved inspiration | Yes | No |
AI creative
| Static ad generation | Yes | No |
| Ads Cloning workflow | Yes | No |
| Creative briefs | Yes | No |
| Reference-image workflow | Yes | No |
Pricing
| Free Tier | 7-day free trial, full access | Completely free (open source) |
| Entry paid tier | $29/month (1 account, includes MCP) | $0 (no paid tier exists) |
| Hidden costs | None. Flat monthly pricing. | Google Cloud project costs (if using Cloud Run), developer time for setup and maintenance |
| MCP call limits | Unlimited | Unlimited (self-hosted) |
More than a data connection
Research
Find what's working right now
Browse 300,000+ ads or import your competitors. Filter by vertical so you see businesses like yours.
- Competitors tracked across Meta, Google, and LinkedIn automatically
- See which ads are evergreens vs. tests your competitors already killed
- Weekly digest in your inbox. No more scrolling ad libraries.
With your agent
Frequently asked questions
Ready to do more than read data?
Track every competitor ad. Search 500,000+ ads. Generate creative in AdKit Studio. Create and manage campaigns across Meta, Google, and TikTok through a draft-first MCP. Two-minute setup.