Loading connector details…
Loading connector details…
Choose a unique username to continue using AgentHotspot
by root-signals • Search & Retrieval
An MCP server that exposes Root Signals evaluators as tools for AI assistants and agents to evaluate, score, and improve LLM responses.
Automatically evaluate and score LLM responses (e.g., clarity, conciseness, faithfulness) and receive justifications for improvement.
Run RAG-aware evaluations using retrieval context to measure faithfulness and relevance against provided contexts.
Enforce coding policy adherence and run multi-evaluator "judge" workflows to combine multiple quality checks into a single decision.
This project bridges the Root Signals API and Model Context Protocol (MCP) clients, letting agents discover and run evaluators (e.g., clarity, faithfulness, coding policy adherence) over model outputs. It supports SSE and stdio transports for integration with MCP-capable clients like Cursor and includes tools to list evaluators/judges and run evaluations by ID or name. Benefits include automated, repeatable quality checks, RAG-aware evaluations with context, and an LLM-as-a-judge workflow to orchestrate multiple evaluators.
Scores are informational only and provided “as is” without warranty. AgentHotspot assumes no liability for actions taken based on these ratings.