searcherries-mcp

# Searcherries MCP server

Searcherries is MCP for GEO. It collects your AI visibility data, connects your search and traffic data, and makes all of it available to GitHub Copilot in VS Code, Claude, Claude Code, Codex, Cursor or any other MCP client. Instead of studying another dashboard, you ask your AI assistant how your site performs and where to act next, and it answers from your own numbers.

Every day, [Searcherries](https://searcherries.com) checks how AI platforms such as ChatGPT, Perplexity, Google AI Overviews, Gemini, Claude and Grok answer the customer questions you track. It records whether your brand is mentioned, which competitors are recommended instead, and which pages get cited. Google Search Console, Bing Webmaster Tools and Google Analytics 4 connect in a few clicks, so your AI visibility sits next to your rankings, queries and AI referral traffic.

Through the MCP connection, your AI assistant can find the questions you lose to competitors, optimize the pages AI cites, expand existing content, close intent gaps, spot new content opportunities and prioritize what to do first. In an editor, it can then change those pages in the same session.

This repository documents the hosted server. The server itself runs at `https://app.searcherries.com/mcp/searcherries`; there is nothing to install or run locally.

## Server

| | |
|---|---|
| Endpoint | `https://app.searcherries.com/mcp/searcherries` |
| Transport | Streamable HTTP |
| Authentication | OAuth 2.1 with dynamic client registration; personal access tokens as a fallback |
| Access | Read-only, limited to the projects of the account you sign in with |
| Registry name | `com.searcherries/mcp` |

## Requirements

A Searcherries account on a paid plan (an active trial counts) with at least one project. [Create an account](https://app.searcherries.com/register).

## Connect

### VS Code and GitHub Copilot

[Install in VS Code](https://insiders.vscode.dev/redirect/mcp/install?name=searcherries&config=%7B%22type%22%3A%22http%22%2C%22url%22%3A%22https%3A%2F%2Fapp.searcherries.com%2Fmcp%2Fsearcherries%22%7D), or add the server to `.vscode/mcp.json` in your workspace:

```json
{
  "servers": {
    "searcherries": {
      "type": "http",
      "url": "https://app.searcherries.com/mcp/searcherries"
    }
  }
}
```

Start the server from the MCP view. VS Code opens the Searcherries sign-in page in your browser; approve access and the tools appear in Copilot Chat in agent mode.

### Other clients

| Client | Setup |
|--------|-------|
| Claude Code | `claude mcp add --transport http searcherries https://app.searcherries.com/mcp/searcherries`, then `/mcp` to sign in |
| Codex | `codex mcp add searcherries --url https://app.searcherries.com/mcp/searcherries`, then `codex mcp login searcherries` |
| Cursor | Add `{"mcpServers": {"searcherries": {"url": "https://app.searcherries.com/mcp/searcherries"}}}` to `~/.cursor/mcp.json` |
| Claude | Add a custom connector with the endpoint URL |

Step-by-step guides for every client: [searcherries.com/docs/mcp/overview](https://searcherries.com/docs/mcp/overview). Clients that cannot complete an OAuth flow can use a personal access token: [searcherries.com/docs/mcp/access-tokens](https://searcherries.com/docs/mcp/access-tokens).

## Use

Ask your assistant in plain language, for example:

- "What should I work on next for my Searcherries project?"
- "Which customer questions do AI assistants answer without naming us, and which competitors do they name instead?"
- "Which pages of this site should we update first? Show the evidence."
- "Write a brief for our pricing page from the search queries and AI questions it should answer."

## Tools

All tools are read-only.

| Tool | Returns |
|------|---------|
| `get-action-plan` | Prioritized next steps for a project, with dated evidence |
| `list-projects` | Your Searcherries projects |
| `get-project-summary` | Compact snapshot: questions, KPIs, competitors, cited domains, 30-day search and traffic totals |
| `get-ai-answers` | Stored AI answers with judgments: how prominently the brand is named, sentiment, off-topic answers |
| `get-competitors` | Brands the AI platforms name, one brand in detail, monthly history |
| `get-citations` | URLs and domains the answers cite, including your own pages |
| `get-search-performance` | Search Console and Bing data: top queries and pages, opportunities, question-shaped searches, trends |
| `get-ai-traffic` | GA4 sessions from AI assistants, landing pages, history |
| `get-project-data` | Bulk sections for deeper analysis |

Full reference: [searcherries.com/docs/mcp/tools](https://searcherries.com/docs/mcp/tools).

## Prompts

The server also ships ready-made workflows as MCP prompts:

| Prompt | Result |
|--------|--------|
| `action-plan` | Up to five next steps with evidence |
| `geo-audit` | Full audit of AI visibility, competitors, sources, search and AI traffic |
| `content-gaps` | Pages to publish or update |
| `competitor-gap` | Comparison with one competitor and a counter-content plan |
| `page-brief` | Brief for one existing or planned page |
| `question-review` | Review of the customer questions you track |
| `weekly-report` | One-page status report |

## Data and privacy

The server cannot change projects, questions, connections or billing. It returns only data that belongs to the Searcherries account you authorize, and it reads stored data: a tool call never triggers a request to Google, Bing or an AI platform on your behalf.

## Support

[Open an issue](https://github.com/eduardmur/searcherries-mcp/issues) or write from your account at [app.searcherries.com](https://app.searcherries.com).