agrag-llm-judge

Features and purpose

Run headless AgRAG prompts in a shared thread and produce an LLM-as-judge evaluation report. Use for evaluating retrieval quality, grounding, graph paths, and tool usage of the agrag agent in headless mode, especially when you need a consistent context window via --thread-id.

Quick start

Follow the repository instructions for this skill and your agent. Review the files and dependencies before adding it to your workflow.