Why FluxLoop?
Building trustworthy AI requires systematic testing and evaluation. FluxLoop provides the foundational tooling.
🎯 Simple Decorator-Based Setup
Instrument existing agent code with minimal changes—just add @fluxloop.agent() and you're tracing. Works with any Python framework or custom agent implementation.
🔄 Argument Replay System
Record complex function arguments (WebSocket callbacks, session data, etc.) from staging, then replay them locally with different content. No manual mocking required.
🧪 Offline-First Simulation
Run experiments on your machine without cloud dependencies. Generate structured artifacts that work with any evaluation backend.
📊 Structured JSON Output
Every simulation produces reproducible, auditable artifacts: summary stats, per-trace records, detailed execution traces, and observation streams.
🚀 CLI Orchestration
Define complex experiments in YAML, generate input variations with LLM, and run batch simulations—all from the command line.
🔌 VSCode Extension
Manage projects, generate inputs, run experiments, and explore results—all from your IDE with visual project management and configuration editing.