# ICONIC — MCP Selection Test for MCP publishers

> ICONIC — MCP Selection Test for MCP publishers is a paid MCP server for agent benchmarks. Measures how often AI models pick your MCP server's tools over rivals' and suggests. It runs as a hosted remote server, so there is nothing to install locally.

- **Kind:** MCP server
- **Pricing:** Paid
- **Starting price:** $49 a month
- **Platforms:** API
- **Website:** https://iconic.onl/agents/selection-test/
- **MCP Registry name:** io.github.pbchrist/iconic-agent-marketplace
- **Categories:** [agent benchmarks](https://ai-powered.com/categories/agent-benchmarks)
- **Listing updated:** 2026-11-18

## About

The service runs your public MCP tool list against same-category servers from the official MCP Registry and checks which ones real models choose for plain-language requests. The report gives a pick rate with a 95% range, overall and per model, plus which rivals won and what the models said was unclear in your descriptions.

After the first run, it rewrites the tool descriptions and tests again on requests the rewrite never saw, so you can compare before and after figures and paste in the new text. The test only reads the tools/list endpoint and never invokes your tools or needs a login.

It is aimed at MCP server publishers who want evidence of how assistants choose between tools. The site states that it cannot see installs, client rankings or real traffic.

## Features

- **Pick rate measurement:** Counts how often three model families chose your server's tools across 20 requests run five times each, with tool order shuffled.
- **Rival server comparison:** Places same-category servers from the official MCP Registry in the same tool list as yours.
- **Read-only tool listing:** Reads only the public tools/list and never calls any tool on your server or others.
- **Before and after rewrites:** Produces rewritten tool descriptions and measures them on held-out requests to show the change.
- **Per-model breakdown:** Shows pick rates per model along with the rivals that won and feedback on unclear descriptions.

## Good for

- Checking whether assistants pick your MCP tools over competing servers
- Finding which rival servers win the same plain-language requests
- Rewriting unclear tool descriptions and measuring the difference
- Deciding whether to revise MCP tool names and descriptions before launch

## Install

### Claude Code (Terminal)

```sh
claude mcp add --transport http iconic-agent-marketplace https://patrick-beastmaster.tailf32530.ts.net/marketplace/mcp
```

### Claude Desktop (Settings → Connectors → Add custom connector)

```sh
https://patrick-beastmaster.tailf32530.ts.net/marketplace/mcp
```

### Cursor (~/.cursor/mcp.json)

```json
{
  "mcpServers": {
    "iconic-agent-marketplace": {
      "url": "https://patrick-beastmaster.tailf32530.ts.net/marketplace/mcp"
    }
  }
}
```

### VS Code (.vscode/mcp.json)

```json
{
  "servers": {
    "iconic-agent-marketplace": {
      "type": "http",
      "url": "https://patrick-beastmaster.tailf32530.ts.net/marketplace/mcp"
    }
  }
}
```

### Windsurf (~/.codeium/windsurf/mcp_config.json)

```json
{
  "mcpServers": {
    "iconic-agent-marketplace": {
      "serverUrl": "https://patrick-beastmaster.tailf32530.ts.net/marketplace/mcp"
    }
  }
}
```

### Other clients (your client's MCP settings (most use this shape))

```json
{
  "mcpServers": {
    "iconic-agent-marketplace": {
      "type": "http",
      "url": "https://patrick-beastmaster.tailf32530.ts.net/marketplace/mcp"
    }
  }
}
```

## FAQ

### How much does the test cost?

The listing states $49 for one server and one report, delivered automatically, usually within an hour.

### Does the test call my MCP tools?

No. It only reads the public tools/list and states that no tool on any server is invoked.

### What happens if no rivals are found?

If at least three rival servers cannot be found, the test does not run and the buyer is refunded.

### Does it need a login to my server?

No login is needed, according to the site, which only requires the MCP URL.

### Is ICONIC — MCP Selection Test for MCP publishers free?

No. ICONIC — MCP Selection Test for MCP publishers is a paid product. Paid plans start at $49 a month.

### How do I add ICONIC — MCP Selection Test for MCP publishers to Claude Code?

Run this command in your terminal: claude mcp add --transport http iconic-agent-marketplace https://patrick-beastmaster.tailf32530.ts.net/marketplace/mcp

## Where this comes from

- Description: written from the project's website and its MCP Registry entry.
- Install steps: from the official MCP Registry entry.

## Alternatives

- [Trialground](https://ai-powered.com/mcp/trialground): Trialground is an MCP server for agent benchmarks.

---
Source: https://ai-powered.com/mcp/iconic-mcp-selection-test-for-mcp-publishers
