From the Benly team. AdSpy.sh and Get MCP Ads are products built by the team behind Benly. Agentik is our exclusive partner bringing both together on a pay-per-use basis.
An AI assistant can explain what CPA means without knowing why your CPA changed. To investigate the second question, it needs access to the right account, a precise date range, the underlying campaign records and enough context to compare like with like. Otherwise, the conversation stops at general advice.
We built Get MCP Ads to connect that conversation to the advertising and analytics data you authorize. It comes from the team behind Benly and provides a hosted MCP gateway for compatible assistants. You bring the account access; the gateway makes supported reporting and operations available as defined tools. Our exclusive Agentik partnership also brings those tools together with AdSpy.sh through a pay-per-use route.
Why we built a bridge to your own accounts
A weekly marketing review often involves copying figures from several interfaces, checking what each metric includes and rebuilding the same context in a document. The first question produces another: which campaigns explain the change, which creatives contributed and whether the landing page or the traffic mix moved too. Each follow-up sends the analyst back to the source.
Get MCP Ads lets that investigation continue in an assistant that can request the relevant records. The purpose is to reduce the gap between a question and its evidence. A useful answer should preserve the platform, account, period and metric context so a teammate can understand what was compared and check the figures against the source.
This is a different job from public competitor research. The assistant is working with your authorized business data, so account selection, permissions and the distinction between reading and changing something belong at the center of the product. Those boundaries are part of making the workflow practical for a single advertiser as well as an agency with several client accounts.
Connect a source, then choose the accounts
Get MCP Ads uses a hosted endpoint, so the client connects to a remote service rather than running a local gateway process. The public setup flow is to connect a platform, select the accounts to share and add the MCP connection to your assistant. The endpoint published by the product is https://mcp.getmcpads.com/mcp.
Platform access and account scope are separate choices
A platform identity may have access to many accounts. That can be convenient for a person working across clients, but it should not automatically become the scope of an assistant’s queries. Get MCP Ads publishes an explicit account selection model: accounts begin unselected, and you choose the ones the assistant can use. Connecting an agency login therefore does not mean sharing every client under that login.
Use the capabilities of your actual connection
The product documents advertising and analytics sources including Meta Ads, Google Ads, TikTok Ads, Google Analytics 4 and Search Console, alongside other integrations. The available tools, release status and write operations differ by source. Begin by asking the assistant which sources and accounts are connected, and what operations are available for the selected account.
Assistant support matters too. Text-based tool use and interactive views are separate capabilities. Creative galleries and comparison experiences require an MCP Apps-compatible client, and account or plan requirements can affect setup. Follow the guide for the client you actually use before making a workflow depend on a particular visual interface.
Investigate a performance change with comparable evidence
Start with an explicit question. “Why did cost per purchase increase last week?” is more useful when it includes the account, comparison periods and conversion definition. Ask the assistant to state those assumptions before it explains a change. A seven-day comparison should use comparable coverage, and the source’s attribution settings belong beside the interpretation.
Then move from the aggregate to the components. Did spend increase, did the recorded purchase count fall, or did both happen? Which campaigns and creative groups account for the difference? A good investigation distinguishes the arithmetic explaining the change from a hypothesis about its cause. More expensive traffic and a weaker conversion rate call for different follow-ups.
Cross-source work needs particular care. Advertising platforms and analytics tools may use different attribution rules, conversion definitions, currencies and reporting time zones. The assistant should retain those distinctions in its output. Several platforms taking credit for the same customer does not create several customers, and a complete-looking table should not conceal a missing source or an incomplete period.
| Question | Evidence to request | Expected output |
|---|---|---|
| What changed this week? | Comparable periods, spend and conversion records | A source-labeled explanation of the movement |
| Which creatives need review? | Creative-level results, exposure and trends | A shortlist with the reason for each review |
| Where is search spend being wasted? | Supported search-term and conversion data | Specific terms and the spend behind each suggestion |
| What should we change? | The evidence, proposed operation and preview | A reviewable decision before any confirmed write |
Bring the creative into the performance conversation
A row of metrics can identify an ad worth reviewing without explaining its execution. Get MCP Ads describes creative galleries, side-by-side comparisons, analysis and daily trends so the team can inspect the creative behind the numbers. Available features depend on the access route, plan, connected source and client support, so begin with the capabilities of your actual workspace.
The useful question is more specific than “Which ad is best?” Ask which creatives contributed meaningful spend and conversions in the chosen period, which need more exposure before a decision and which warrant a closer fatigue review. Creative performance depends on the audience, placement and delivery context as well as the asset. A brief should carry that context forward.
Business and landing-page reviews extend the investigation beyond the ad. If a campaign promises simplicity but sends visitors to an unclear offer, changing the opening hook may miss the larger problem. Reviewing the creative, account evidence and destination together helps a team decide which part of the journey deserves a test. A review should keep observations separate from causal claims that the available data cannot establish.
Preview supported changes before confirming them
Reporting and execution are related, but a conversation should have a clear boundary between them. Get MCP Ads describes a preview-first model for supported writes: a tool returns the proposed change, and applying it requires a second, confirmed call. The product also states that newly created campaigns start paused. A proposed change is therefore something you can inspect before it becomes an account operation.
Review the account and platform first, then the object being changed, the old and new values, and any schedule or currency details. “Increase the budget” is incomplete without a specific campaign and amount. If the preview identifies the wrong campaign, correct the request before confirming. If a platform or tool does not support the intended operation, keep the output as a recommendation rather than assuming the change happened.
After confirmation, inspect the result and the available change history. A generated recommendation, a preview and an applied change are three different states. Keeping them distinct makes it possible to explain what the assistant suggested, what a person authorized and what the platform actually accepted.
Pair your own evidence with AdSpy’s competitor context
Your account data can tell you which creatives deserve attention. It cannot, by itself, show every argument competitors are presenting to the market. That is why we built two complementary products. Get MCP Ads examines the accounts you authorize; AdSpy.sh researches public competitor creative.
Imagine that a review identifies a group of ads worth refreshing. AdSpy can help inspect how relevant brands demonstrate a similar product, address an objection or introduce an offer. Those observations can inform a new brief. Get MCP Ads can later help examine the result in your own accounts, with the account and attribution context preserved.
Read our AdSpy.sh story on competitor creative intelligence for a deeper explanation of research sampling, Brand X-Ray and collections. The combination works because the evidence remains identifiable: a competitor example supports a creative hypothesis, while your own data supports a statement about your campaign performance.
How Agentik makes the combination available on pay per use
Our exclusive Agentik partnership brings AdSpy.sh and Get MCP Ads together through metered tool access. Agentik’s catalog lists Get MCP Ads operations for connections, account reporting, creative workflows, business reviews, landing-page reviews and supported changes. The assistant can discover the relevant operation and inspect its current price before running it.
As checked on October 22, 2026, that catalog includes free connection and supporting tools, account snapshots listed at $0.005 per call, and supported confirmed campaign changes listed at $0.05 per change. Other operations have different rates or usage-based pricing. These are individual tool examples, not a quote for an entire review. A multi-step question can consume several calls, and Agentik’s top-up fees are separate from tool pricing.
You still need your own Get MCP Ads connection, made once per workspace, and the appropriate account authorization. Paying through Agentik does not replace those permissions, include the advertising platform’s media budget or imply that every direct product feature is exposed by the same tool. Read the complete Agentik partnership guide for billing, spending caps and the combined workflow.
Start with one account and one real question
Choose a platform, select the intended account and ask for a bounded review of a recent period. Have the assistant name its sources and comparison assumptions. Follow up on one meaningful change, inspect the creative or campaign evidence behind it and decide whether the next step is more investigation, a new brief or a preview of a supported operation.
We built Get MCP Ads so those follow-ups can stay connected to real account evidence. Explore Get MCP Ads and its setup guides to choose your client and access route. Whether you use the product directly or through Agentik, begin with a question whose answer you can check against the account you connected.
Product references checked October 22, 2026: Get MCP Ads, its published security model and its Agentik tool catalog. Check current source status, permissions and pricing before planning an operation.
Explore the rest of the ecosystem
Read the companion stories to see how competitor research, your own advertising data and access through Agentik fit together.
- Agentik: Pay-Per-Use Tools for AI Assistants — How our exclusive Agentik partnership brings AdSpy.sh and Get MCP Ads to AI assistants, with metered tools, one prepaid balance and clear spending controls.
- AdSpy.sh: Competitor Ad Intelligence and Creative Research — Inside AdSpy.sh, built by the team behind Benly: competitor ad search, creative analysis, Brand X-Ray and shared research, also available through Agentik on pay per use.
