# AdsGPT [Health: Active]

**Category:** 💻 Developer Tools  
**Repository:** https://github.com/wortise/agent-plugins  
**GitHub Stars:** 0  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/adsgpt

## Description
Create, review and optimize your Google Ads campaigns, approving every change yourself.

## Claude Desktop Quick Installation
No confirmed install details yet, so no config is published. Follow the setup steps in the repository above.

## Documentation

## What AdsGPT MCP server does

The AdsGPT MCP server supports campaign work across advertising platforms connected to AdsGPT, such as Google Ads. Its stated workflow covers three activities: creating campaigns, reviewing existing campaign work, and optimizing campaigns. The intended control model keeps approval with the user, so an agent can prepare or suggest changes without removing the human decision step.

AdsGPT is one plugin in Wortise’s agent-plugin repository. The repository contains separate plugin directories, client manifests, and an MCP server registry entry. The AdsGPT registry identifier is `dev.adsgpt/adsgpt`.

## How it works

The AdsGPT MCP server is packaged as an agent plugin rather than presented only as a standalone repository integration. Client-specific manifests describe how the plugin is listed and installed, while the registry metadata describes its MCP server for discovery through the official MCP Registry.

The parent repository acts as a marketplace for Wortise plugins. An agent client installs the AdsGPT plugin from that marketplace, then uses the plugin’s MCP server for campaign-related work. The available ad platforms depend on what is connected to AdsGPT; Google Ads is given as an example, not as the only supported platform.

The documented workflow emphasizes review and approval. This makes the server relevant when an advertising operator wants an AI agent to help prepare, inspect, or improve campaign changes while retaining control over whether those changes are approved.

## Setup and configuration

Installation is performed through the Wortise marketplace or a client’s plugin system. Claude Code uses the marketplace command followed by the `adsgpt@wortise` plugin identifier. Claude web and desktop users add `wortise/agent-plugins` under Customize > Plugins. Codex users add the marketplace with `codex plugin marketplace add wortise/agent-plugins`, then install a plugin from Wortise through `/plugins`.

The repository also documents marketplace support for Cursor and Grok. Cursor has a plugin manifest at `.cursor-plugin/plugin.json`, and the repository includes a corresponding marketplace manifest. The README says that each plugin’s own README contains its runtime, data sent and read, and sign-in instructions; those details are not included in the supplied material, so authentication requirements cannot be specified here.

## Tools and capabilities

The AdsGPT MCP server is described as providing tools to:

- Create advertising campaigns.
- Review campaign changes or configurations.
- Optimize advertising campaigns.
- Work with ad platforms connected to AdsGPT, including Google Ads.

The supplied documentation does not name individual tool identifiers or describe campaign fields, account selection, reporting operations, or platform-specific actions.

## Limitations and notes

The AdsGPT MCP server’s supported platforms are described only as platforms connected to AdsGPT, with Google Ads as an example. The material does not establish that every advertising platform is supported or provide a complete platform list.

Approval is part of the stated workflow: users approve every change themselves. The available excerpt does not explain whether approval is implemented as a particular MCP tool, a client confirmation step, or another mechanism.

No environment variables, command-line runtime options, API scopes, or sign-in procedure are specified in the supplied documentation. Consult the AdsGPT plugin README for those details before deployment. The repository is licensed under MIT.

_Full upstream README: https://allmcps.com/mcp/adsgpt/readme_