CLI entry point and commands for the RAG evaluation toolkit
TypeScript execution environment and REPL for node.js, with source map support
Core types, schemas, and domain models for RAG evaluation
evaluate statically-analyzable expressions
MCP server exposing RAG evaluation tools (judge, suite, gate)
Cost tracking, pricing, budgeting, and reporting for RAG evaluations
RAG evaluation metric scorers: faithfulness, relevance, context precision/recall
Dataset loading, validation, generation, and versioning for RAG evals
Structured logging, OpenTelemetry tracing, and metrics for RAG evaluations
LLM-as-judge with calibration, consensus voting, and cost tracking
Quality gates and CI/CD regression checks for RAG evaluations
Central evaluation orchestrator that ties metrics, judge, cost, gate, and dataset together
The grunt command line interface
Simple JavaScript expression evaluator
Evaluate node require() module content directly
Mathematical expression evaluator fork with exports map, prototype pollution and code injection security fixes
A flexible math expression evaluator
require or eval modules
Structured logging, tracing, and metrics for hybrid RAG systems
Core domain types, schemas, and shared utilities for hybrid RAG systems
Document loading, preprocessing, and chunking strategies for hybrid RAG systems
Mark scopes for deopt which contain a direct eval call
JavaScript expression parsing and evaluation.
Alias for eval global.