Features
Multi-turn agent simulations with LLM-powered user simulator
Per-turn judge agent with pass/fail criteria
Adversarial red-teaming (Crescendo escalation, refusal detection)
Voice agent testing with latency metrics and noise injection
Tool-call verification across long dialogues
Framework-agnostic adapters (LangGraph, CrewAI, Pydantic AI, etc.)
Run locally or in CI/CD (pytest/vitest)
Visual debugger for simulation replay
Pause, evaluate & annotate mid-conversation
Open-source Scenario SDK (Python + TypeScript, MIT)
Langy: AI assistant to generate test plans from plain-English goals
Prompt Registry with versioning and A/B testing
AI Governance: virtual keys, budgets, routing policies, audit trail
Event sourcing engine for real-time processing of traces, evals, simulations
Self-host with ClickHouse (no Elasticsearch required)
Local model execution for offline inference
Connect online APIs: OpenAI (ChatGPT), Anthropic (Claude), Google (Gemini), Meta (Llama), Mistral AI, Alibaba (Qwen),
123+ HuggingFace models available
Memory feature (coming soon) for context persistence
Minimalist UI with dark mode
Open source under MIT license on GitHub
Cross-platform desktop app (Mac, Windows, Linux)
Version 0.8.3 with bug fixes and performance improvements
