CompletionKit

Overview

CompletionKit is a Model Context Protocol (MCP) server: Prompt evals over MCP: run a prompt on your dataset, score each output 1-5 with an LLM judge. Because it speaks the Model Context Protocol, AI assistants can use CompletionKit as a set of callable tools instead of you copy-pasting between apps.

It's a hosted (remote) server — point your client at https://completionkit.com/mcp and you're connected, with nothing to install locally. The current release is v1.0.0, with source at homemade-software-inc/completion-kit on GitHub.

On mcp.site, CompletionKit sits in the AI & Agents category. Any MCP client can use it — Claude Desktop, Claude Code, Cursor, VS Code, Windsurf, Zed among them — via the install snippets below. If you maintain CompletionKit for completionkit.com, claim this listing to verify ownership, earn the Verified badge, and keep the details current.

Installation

Settings → Connectors → Add custom connector
Name: CompletionKit
URL:  https://completionkit.com/mcp