1. The endpoint
Point your client at the MCP server below. It speaks the Model Context Protocol over HTTPS (JSON-RPC 2.0), protocol version 2025-06-18.
https://api.ovaideu.com/api/mcpOvaideu runs a Model Context Protocol (MCP) server, so an AI agent such as Claude, ChatGPT, Codex or Gemini can run your Google Ads, GA4, Search Console, Tag Manager, Business Profile and Meta Ads work in plain language — inside the same tenant-scoped, user-bound access as the dashboard. Add one connector URL, authenticate with your own API key, and you are set.
MCP connector URL
https://api.ovaideu.com/api/mcpAlso reachable at https://ovaideu.com/mcp — same server.
Once connected, your agent can read from and make authorized changes across the accounts you authorize — nothing wider. Every write is scoped to your key, logged, and read back after it runs.
Google Ads — build, edit and optimize real campaigns
Google Analytics 4 — key events and conversions
Google Search Console — queries and top pages
Google Tag Manager — tags and triggers
Google Business Profile — profile and locations
Meta Ads — campaigns, ad sets and reporting
Whatever client you use, the connection is always the same URL, the same header, and a key you create yourself. Get these right first — the per-client steps below just plug them in.
Point your client at the MCP server below. It speaks the Model Context Protocol over HTTPS (JSON-RPC 2.0), protocol version 2025-06-18.
https://api.ovaideu.com/api/mcpSend your Ovaideu API key as a Bearer token. Web connectors and the desktop bridge do this for you; for direct API calls, add the header yourself.
Authorization: Bearer YOUR_OVAIDEU_API_KEYSign in to Ovaideu, activate your own MCP access, and create a user-bound key in the dashboard. Workspace admins can also manage internal users from the same page.
Claude and ChatGPT can connect to the remote MCP server directly. Add it once as a custom connector and authorize your own Google or Meta account in your browser.
Paste this into either client:
https://api.ovaideu.com/api/mcpClients that speak MCP over stdio reach the server through the bundled Node bridge. It forwards each call to the same URL with your key as a Bearer token — you never paste a secret into a web form.
Sign in to the dashboard, open MCP Management, activate your MCP access, and create one scoped key bound to your own user email.
Download the bridge package, place ovaideu-desktop-mcp-bridge.mjs on your computer, and note its absolute path.
Paste the Claude JSON or Codex TOML below into your client and set the bridge path. Requires Node.js on the machine.
Replace PASTE_YOUR_OVAIDEU_API_KEY_HERE and user@example.com with your real values, then restart the client.
{
"mcpServers": {
"ovaideu": {
"command": "node",
"args": ["/ABSOLUTE/PATH/TO/ovaideu-desktop-mcp-bridge.mjs"],
"env": {
"OVAIDEU_MCP_URL": "https://api.ovaideu.com/api/mcp",
"OVAIDEU_MCP_ORIGIN": "https://claude.ai",
"OVAIDEU_MCP_TOKEN": "PASTE_YOUR_OVAIDEU_API_KEY_HERE",
"OVAIDEU_USER_EMAIL": "user@example.com"
}
}
}
}[mcp_servers.ovaideu] command = "node" args = ["/ABSOLUTE/PATH/TO/ovaideu-desktop-mcp-bridge.mjs"] [mcp_servers.ovaideu.env] OVAIDEU_MCP_URL = "https://api.ovaideu.com/api/mcp" OVAIDEU_MCP_ORIGIN = "https://chatgpt.com" OVAIDEU_MCP_TOKEN = "PASTE_YOUR_OVAIDEU_API_KEY_HERE" OVAIDEU_USER_EMAIL = "user@example.com"
The same endpoint answers direct HTTP calls. Run a tools/list request with your key to confirm access before wiring up a client. A valid key returns the tools it is allowed to see.
curl -s https://api.ovaideu.com/api/mcp \
-H "Authorization: Bearer YOUR_OVAIDEU_API_KEY" \
-H "Content-Type: application/json" \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'A 401 means the key is missing or wrong; a 403means your signed-in user does not have active MCP access yet — open API & MCP and activate your access.
Use View to inspect a file in the browser, or Download to save the package locally.
Give this PDF to your AI assistant. It includes the connector URL, the authentication step and the shortest path through account authorization.
Start here. A short setup path for desktop MCP users and AI assistants.
Ovaideu follows the open Model Context Protocol, so it works with any client that speaks MCP. Web clients use the connector URL directly; desktop and CLI clients use the bridge.
The primary surface. One remote server, user-bound scoped access, and the same tenant-scoped access as the dashboard.
The same endpoint answers direct HTTPS calls for authorized server-to-server workflows in your own workspace.
A public installer command is planned after MCP/API verification. Until then, use the bridge package and PDF.
Ovaideu workspaces include scoped keys, MCP connectors, usage controls and audit logs after account setup.
No card required · White-label for agencies