Most AI tools require you to leave your workflow, open a browser tab, and context-switch into a dedicated UI. AdsCreator works differently.

AdsCreator exposes an MCP (Model Context Protocol) server — which means you can generate on-brand ad creative directly from Claude Desktop, Cursor, or any other AI tool that supports MCP, without leaving your existing workflow.

This post explains what that means, how to set it up, and what you can build with it.

What Is MCP?

Model Context Protocol is an open standard developed by Anthropic that lets AI assistants connect to external tools and data sources. Instead of pasting content into a chat window and copying results back out, MCP-compatible tools plug directly into your AI assistant's context.

You describe what you want in natural language. The AI calls the tool. The result comes back inline.

Practical example without MCP:

Open browser
Navigate to adscreator.ai
Paste URL
Download creative
Switch back to your workflow

With MCP:

"Generate a Facebook feed ad for acmeco.com focused on their enterprise plan"

→ Ad creative generated inline, in your AI assistant, no tab switching.

Why Developers and AI Builders Use It

AdsCreator's MCP server is particularly valuable for three types of users:

AI workflow builders — If you're building automated marketing pipelines (content → copy → creative → scheduling), AdsCreator's MCP server lets ad generation sit natively inside that pipeline. No API glue code required for the creative step.

Developers with non-technical clients — Build a simple internal tool where a client types a product URL and gets branded ad creative back, all powered by Claude + AdsCreator MCP in the background.

Power users in Claude Desktop — If you already use Claude Desktop for complex tasks, adding AdsCreator MCP means you can generate ads as part of any research, briefing, or strategy session without switching context.

Installing the AdsCreator MCP Server

Prerequisites

Claude Desktop (or another MCP-compatible client)
An AdsCreator API key (available from your dashboard at adscreator.ai)
Node.js 18+ installed

Step 1: Add to Claude Desktop config

Open your Claude Desktop configuration file:

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json Windows: %APPDATA%\Claude\claude_desktop_config.json

Add the AdsCreator server to the mcpServers block:

{
  "mcpServers": {
    "adscreator": {
      "command": "npx",
      "args": ["-y", "@adscreator/mcp-server"],
      "env": {
        "ADSCREATOR_API_KEY": "your-api-key-here"
      }
    }
  }
}

Step 2: Restart Claude Desktop

After saving the config, restart Claude Desktop. You should see the AdsCreator tools appear in the tools panel.

Step 3: Verify connection

In Claude Desktop, ask:

"What AdsCreator tools are available?"

Claude will list the available tools — brand extraction, ad generation, platform sizing, and variant generation.

What Tools Are Exposed

The AdsCreator MCP server exposes four core tools:

`extract_brand`

Extracts brand DNA from any URL — colors, fonts, messaging, tone, visual style.

Input: { "url": "https://example.com" }
Output: Brand profile (colors, typography, value prop, tone)

`generate_ad`

Generates ad creative using a brand profile and campaign parameters.

Input: {
  "brand_url": "https://example.com",
  "platform": "meta" | "linkedin" | "google_display" | "tiktok",
  "format": "feed" | "stories" | "carousel" | "banner",
  "objective": "awareness" | "consideration" | "conversion",
  "headline": "optional override",
  "copy": "optional override"
}
Output: Image URL, dimensions, copy used, brand tokens applied

`generate_variants`

Generates multiple creative variants from a single brand + brief. Useful for A/B testing setups.

Input: {
  "brand_url": "https://example.com",
  "platform": "meta",
  "count": 5,
  "vary": ["hook", "layout", "copy_angle"]
}
Output: Array of variant image URLs with metadata

`resize_for_platform`

Takes an existing creative and resizes/reformats it for a different platform.

Input: {
  "source_image_url": "...",
  "target_platform": "tiktok",
  "target_format": "in_feed"
}
Output: Resized image URL at correct dimensions

Example Workflows in Claude Desktop

Workflow 1: Campaign Brief to Creative

You: "I'm launching a campaign for stripe.com targeting SaaS founders.
      Generate 3 Meta feed ad variants focused on their payment infrastructure."

Claude: [calls extract_brand("stripe.com")]
        [calls generate_variants({brand_url, platform: "meta", count: 3, vary: ["copy_angle"]})]
        → Returns 3 on-brand creative variants, downloads available

Workflow 2: Multi-Platform Asset Generation

You: "Generate the full ad creative set for shopify.com's new B2B features —
      need Meta feed, LinkedIn single image, and Google display banner."

Claude: [calls extract_brand once]
        [calls generate_ad three times in parallel for each platform]
        → Returns complete cross-platform creative set

Workflow 3: Agency Client Onboarding

You: "New client: acmeco.com. Extract their brand profile and generate
      a starter pack: 2 Meta variants, 1 LinkedIn, 1 Google Display."

Claude: [extracts brand profile]
        [generates 4 assets across platforms]
        → Full starter creative set, ready for client review

Building Automated Pipelines

The MCP server is designed to be called programmatically, not just interactively. Here's a sketch of what an automated pipeline looks like using the Claude API with tool use:

import Anthropic from "@anthropic-ai/sdk";

const client = new Anthropic();

async function generateCampaignCreative(brandUrl: string, campaign: Campaign) {
  const response = await client.messages.create({
    model: "claude-opus-4-6",
    max_tokens: 1024,
    tools: adscreatorTools, // MCP tools registered as Claude tools
    messages: [
      {
        role: "user",
        content: `Generate a complete creative set for ${brandUrl}.
          Campaign objective: ${campaign.objective}
          Platforms: ${campaign.platforms.join(", ")}
          Key message: ${campaign.keyMessage}`,
      },
    ],
  });

  // Handle tool calls, collect generated assets
  return extractAssets(response);
}

This pattern lets you build internal tools, client portals, or fully automated creative generation workflows — with AdsCreator's brand-awareness built in at the generation step.

What Makes This Different From the API

AdsCreator also offers a REST API for direct integration. The MCP server differs in two important ways:

Natural language interface — MCP tools are invoked by AI models in response to natural language requests. You don't need to structure API calls; you describe what you want and the model handles the tool orchestration.

Workflow composability — Because MCP tools run in the same context as your AI assistant, they can be combined with other tools in a single conversation. Generate a brand profile, write copy with a different tool, generate creative, then draft a campaign report — all in one session.

The API is better for high-volume programmatic generation. MCP is better for workflow integration and AI-assisted campaign production.

Browse Ad Examples

See what the MCP server can generate:

Get Started

Grab your API key from adscreator.ai
Add the MCP server config to Claude Desktop
Restart and start generating ads in natural language

The full MCP server documentation, including all tool schemas and example prompts, is available at adscreator.ai/developers.

Key Takeaways

MCP eliminates context switching — generate ads from inside Claude Desktop or Cursor without leaving your workflow
Four core tools — extract brand, generate ad, generate variants, resize for platform
Config takes 2 minutes — one JSON block in your Claude Desktop config and you're running
Pipeline-ready — the MCP server is designed to be called programmatically for automated creative workflows
Natural language interface — describe what you want; the AI handles tool orchestration