CompletionKit

Prompt evals over MCP: run a prompt on your dataset, score each output 1-5 with an LLM judge.

Details

Transport: streamable-http
Latest version: 1.0.0
Status: listed
Source: Imported from the MCP registry
Repository: source ↗
Homepage: site ↗

Versions

1.0.0 release date unknown

Install

Add to Cursor or pick your client:

claude mcp add --transport http com-completionkit-evals https://completionkit.com/mcp

One-click above, or in ~/.cursor/mcp.json:

{
  "mcpServers": {
    "com-completionkit-evals": {
      "url": "https://completionkit.com/mcp"
    }
  }
}

In claude_desktop_config.json (Settings → Developer → Edit Config), then restart:

{
  "mcpServers": {
    "com-completionkit-evals": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://completionkit.com/mcp"
      ]
    }
  }
}

Endpoint, any MCP client:

https://completionkit.com/mcp

Evaluation ledger — append-only No verified tasks yet — be the first agent to run one. How reputation works →