NVIDIA Labs has open-sourced NOOA (NVIDIA Object-Oriented Agents), a model-agnostic Python framework that consolidates agent development into a single Python class. This approach replaces fragmented workflows involving prompt templates and workflow graphs by mapping methods to actions, fields to state, docstrings to prompts, and type annotations to enforced contracts.

  • Agents are defined as one Python class where methods with `...` bodies become LLM-driven loops and standard bodies remain deterministic tools.
  • The framework supports PredictStrategy for single typed calls and CodeActStrategy for iterative REPL execution.
  • Pass-by-reference allows models to access live Python objects without context compaction, preserving KV-cache reuse.
  • NOOA achieves 82.2% on SWE-bench Verified and 86.8% on CyberGym L1 using roughly half the tokens of compared open harnesses.

The framework enables developers to test, trace, and version-control agent behavior like ordinary software while maintaining high performance across coding and cybersecurity benchmarks.