Symbolic computing and numeric evaluations for JavaScript and Node.js

Evaluations - Open infrastructure for AI agents

A library for running evaluations for AI use cases

Universis evaluations library

Evaluate operations in a scope while retaining resources across evaluations

JavaScript client for running remote evaluations with Scoundrel.

Policy information point service for supplying context facts to evaluations.

Command-line interface for Droyd research competitions and evaluations

Zero-config Node.js wrapper for M365 Copilot Agent Evaluations CLI (Python-based Azure AI Evaluation SDK)

EvalGuard CLI - Run LLM evaluations and security scans from the command line

Minimal TypeScript library for running LLM-as-a-judge evaluations

Command-line interface for SwarmX workflows and evaluations.

Reward functions and trace helpers for Ignition Agent Trainer evaluations.

Bounded deterministic, trace, model-judge, pairwise, and CI evaluations for Prism runs.

Evidence-backed model metadata, evaluations, capability requirements, and routing policy.

Evaluation framework with Effect-based architecture for running parallel AI evaluations

Run LangWatch locally with one command — `npx @langwatch/server`. Complete LangWatch stack: observability, evaluations, AI gateway, agent simulations.

Agent composition layer for identity, doctrine, skills, models, knowledge, permissions, evaluations, and upkeep.

Symbolic computing and numeric evaluations for JavaScript and Node.js

Deliberately minimal EntryPoint 0.7 bundler for Moesi's local Anvil fork evaluations.