Adsroid MCP is an MCP server that connects AI assistants to advertising platforms. Through a single endpoint, it gives any MCP-compatible AI full access to your ad accounts, reading performance data, managing campaigns, building structures, and running experiments, all from a natural language conversation, and without touching a dashboard.
This documentation covers everything you need to connect, understand, and use Adsroid MCP in production.
What is MCP?
MCP stands for Model Context Protocol. It is an open standard developed by Anthropic that defines how AI assistants communicate with external tools and services.
Before MCP, AI models were isolated from the world. They could reason about information you gave them, but they had no structured way to query live data, trigger actions, or interact with external systems. Every integration was custom-built, brittle, and platform-specific.
MCP solves this at the protocol level. It defines a standard interface through which an AI assistant can:
- Discover what tools are available on a connected server
- Understand what each tool does, what parameters it expects, and what it returns
- Call those tools autonomously within a conversation, with full transparency
When you connect an MCP server to Claude, the AI doesn’t just know about your tools, it can use them. It reasons about which tool fits the task, constructs the correct call, interprets the result, and continues the conversation. From the user’s perspective, it feels like talking to someone who has direct access to your systems.
MCP is now natively supported by Claude and is rapidly establishing itself as the standard for connecting AI to real-world data and operations.
What is Adsroid MCP?
Adsroid MCP is an MCP server purpose-built for advertising management. It exposes your advertising accounts as a structured set of tools that any MCP-compatible AI assistant can call in real time, directly against the ad platform’s API.
What sets Adsroid MCP apart is its architecture. Most AI integrations are designed around a single platform, a single account, or a single use case. Adsroid MCP is built around a fundamentally different model: one single endpoint for every platform, every account, and every project.
You connect once at https://mcp.adsroid.com/mcp and from that point, your AI assistant has access to all your advertising accounts across all connected platforms, within a single conversation. No re-authentication. No context switching. No parallel dashboards.
This architecture is what makes Adsroid MCP uniquely powerful as a foundation for AI-driven advertising operations. As new platforms are added to the server, they become immediately available through the same endpoint, without any change on the client side.
Capabilities
Adsroid MCP organises its tools into four distinct capability areas.
Analytics
All read operations fall under Analytics. The AI can query account-wide metrics, campaign and ad group performance, keyword and search term data, audience segments, asset ratings, device and geo breakdowns, hourly trends, conversion actions, budgets, and experiment results. Analytics tools never modify your account — they are strictly read-only.
Campaign Management
Management tools cover write operations on existing elements: updating campaign and ad group statuses, adjusting budgets and bidding strategies, modifying keywords and negative keywords, managing extensions, and controlling A/B experiment states. Every management action requires explicit user confirmation before execution.
Campaign Builder
Builder tools handle the creation of new elements from scratch: campaigns, ad groups, responsive search ads, keywords, budgets, and extensions. Like management tools, all creation actions require confirmation before any change is applied to the account.
A/B Testing
The Experiments module covers the full lifecycle of A/B tests, from creating an experiment and configuring its arms, to launching, monitoring comparative metrics, and graduating or ending the test. Experiment tools span both read and write operations and are documented as a standalone section given their distinct workflow.
How it works
When you send a message to Claude with Adsroid MCP connected, the following happens:
- Claude reads your message and determines which tool, or sequence of tools, is needed to fulfil the request.
- Claude constructs the tool call with the correct parameters and sends it to the Adsroid MCP server.
- The MCP server authenticates the request using your API key, identifies the target project and ad account, and forwards the call to the advertising platform’s API.
- The platform API returns the data or confirms the action. The MCP server passes the result back to Claude.
- Claude interprets the result and continues the conversation, summarising data, flagging anomalies, proposing next steps, or asking for confirmation before applying a change.
Every request is processed in real time. Adsroid does not store, cache, or log your account data at any point in this flow.
Supported platforms
Adsroid MCP is designed to cover every major advertising platform. Connectors are added progressively, all available through the same endpoint without any configuration change.
| Platform | Status |
|---|---|
| Google Ads | ✅ Available |
| Meta Ads | ✅ Available |
| TikTok Ads | Coming soon |
| LinkedIn Ads | Coming soon |
| Microsoft Ads | Coming soon |
| Google Search Console | Coming soon |
| Google Analytics 4 | Coming soon |
| Hubspot | Coming soon |
| WordPress | Coming soon |
Core principles
Zero data retention. Every tool call is a direct, real-time pass-through to the ad platform’s API. No account data is stored or cached on Adsroid’s infrastructure.
Confirmation before action. All write and create operations are designed to require explicit user approval. The AI presents what it intends to do, including the exact parameters, before executing any change. You remain in full control of your accounts at all times.
One connection, all your accounts. A single Adsroid API key covers all projects in your workspace. You can switch between accounts and platforms within the same conversation without reconnecting.
Platform-agnostic by design. The tool architecture is abstracted from any specific platform. As new connectors are released, they integrate into the same endpoint and follow the same capability model, Analytics, Management, Builder, Experiments.
