Codabench Mcp

Overview

Codabench Mcp is a Model Context Protocol server that connects to the Codabench REST API and automates the full ML-benchmark participant workflow. Any MCP-compatible AI assistant can call its tools directly, letting models decide when and how to engage with Codabench during conversations or agent runs.

The server runs locally via the codabench-mcp Python package installed with uvx. It communicates over stdio and requires CODABENCH_API_TOKEN for authentication. The latest release is v0.1.1, distributed under the MIT license with source code available at harshitAgr/codabench-mcp on GitHub.

Codebench Mcp is categorized as Development on mcp.site and integrates with all major MCP clients including Claude Desktop, Claude Code, Cursor, VS Code, Windsurf, and Zed. If you maintain this server, claim your listing to earn the Verified badge and keep the information up to date.

Installation

// claude_desktop_config.json
{
  "mcpServers": {
    "codabench_mcp": {
      "command": "uvx",
      "args": [
        "codabench-mcp"
      ],
      "env": {
        "CODABENCH_API_TOKEN": "<codabench-api-token>"
      }
    }
  }
}

Requires CODABENCH_API_TOKEN