AgentLab Simulator

About

AgentLab Simulator is a ready-to-run, self-hosted QA and red-team platform for conversational AI agents.It connects to AI agents over HTTP, runs repeatable multi-turn simulations, evaluates their behavior, and generates reliability reports with failure evidence and worst-case conversations.Key features:• Automated multi-turn agent simulations• Prompt-injection and jailbreak testing• Refund, angry-customer and policy-compliance scenarios• Language-switching and changing-requirements tests• Generic HTTP and OpenAI-compatible adapters• Deterministic and optional LLM-based evaluation• Severity-weighted reliability scoring• Conversation transcripts and failure evidence• Baseline comparison between agent versions• CI quality gates• Human review and score overrides• Encrypted credentials and tamper-evident audit history• Extensible scenarios, mutations and gradersThe product includes a Next.js dashboard, FastAPI backend, PostgreSQL database, Redis worker queue, Docker Compose deployment, database migrations, automated tests, built-in scenario library, demo agent and complete technical documentation.AgentLab v1.0.0 is feature-complete and Docker-verified. The working browser flow creates a project, connects an agent, runs red-team simulations and produces a reliability report.The release passes 27 backend tests, frontend type checking and linting, production builds and dependency security checks.This is a pre-revenue source-code acquisition. No revenue, customers or traffic are being claimed. The buyer receives the complete product foundation and can rebrand it, launch a hosted SaaS, sell self-hosted licenses or offer managed AI-agent testing services.