Enables building and running custom LLM benchmarks with multi-judge evaluation, supporting GUI, MCP client, and CLI usage for ranked, auditable results.
