A lightweight monitor for tracking LLM output quality, deployed on Cloudflare Workers and accessible via both REST and MCP protocols. The implementation uses streamable HTTP transport, making it straightforward to integrate into existing workflows without heavyweight dependencies. You'd reach for this when you need to programmatically assess or log the quality of responses from language models, whether you're building evaluation pipelines, debugging prompt engineering experiments, or maintaining quality metrics across production LLM integrations. The dual REST and MCP interface means you can query it from traditional HTTP clients or hook it directly into Claude's tool ecosystem.