Nvidia Object-Oriented Agents: An agent is a Python class

NVIDIA-labs released NOOA (NVIDIA-labs OO Agents), a model-agnostic Python framework that represents AI agents as Python classes, unifying state, capabilities, prompts, and typed interfaces in a single object. The framework, available on GitHub, supports familiar Python development workflows and includes optional sub-packages for CLI, memory, and evaluation, but is research software that requires sandboxed execution due to risks from LLM-generated code.

NVIDIA-labs OO Agents NOOA is a model-agnostic Python framework designed to support reliable AI agent development. Many agent frameworks represent prompts, tools, callbacks, and workflows as separate abstractions. NOOA offers an alternative object-oriented interface that brings these concepts together in a Python class. NOOA lets developers express an agent’s state, capabilities, prompts, and typed interfaces through a single Python class: python from nooa import Agent The agent is a Python object. class SupportAgent Agent : """You are a support agent.""" State lives on the object. Fields are typed. order db: OrderDB Ordinary method. Just Python. def is refund eligible self, order: Order - bool: return order.delivered and order.days since delivery <= 30 Agentic method: the runtime hands this to an LLM. async def triage self, message: str, order: Order - Ticket: """Create a typed support ticket.""" ... What's happening here: Agents are Python objects. Fields are state, methods are capabilities, docstrings are prompts, type annotations are contracts.A method with ... bodies are LLM-driven. ... becomes an agentic loop; a real body stays deterministic Python. Code as action. The model acts by writing Python in a Jupyter-style REPL with access to self , imports, and helpers — Python methods and type annotations supply the callable interfaces, reducing the need to write separate tool-schema definitions. Pythonic and agent-ready. Typed I/O with auto-retry, live-object arguments passed by reference, and model-callable context and event APIs — designed around agent-oriented Python workflows. This design supports familiar Python testing, tracing, refactoring, and version-control workflows — just like the rest of your software . Read the paper for the design principles and evaluation results: NVIDIA OO Agents: Native Python Object-Oriented Agents https://arxiv.org/abs/2607.20709 . Install directly from GitHub with uv https://docs.astral.sh/uv/getting-started/installation/ . Add the core framework to a new or existing Python project: uv init my-agent-project cd my-agent-project uv add "nooa @ git+https://github.com/NVIDIA-NeMo/labs-OO-Agents.git@main" Optional sub-packages — CLI, memory, evaluation pipeline All of these live in the same repo and are addressed with subdirectory=… . CLI beta : the nooa command, trace viewer, eval runner uv add "nooa-cli @ git+https://github.com/NVIDIA-NeMo/labs-OO-Agents.git@main subdirectory=packages/nooa-cli" Long-term memory subsystem MemoryManager uv add "nooa-memory @ git+https://github.com/NVIDIA-NeMo/labs-OO-Agents.git@main subdirectory=packages/nooa-memory" Evaluation pipeline for agent testing uv add "eval pipeline @ git+https://github.com/NVIDIA-NeMo/labs-OO-Agents.git@main subdirectory=util/eval pipeline" NOOA is research software , and agents can be configured to execute LLM-generated code. We welcome contributions and fixes, but expect rough edges. LLM-generated code may take dangerous or unwanted actions, incuding sending private data to uncontrolled locations, deleting files, or modifying its environments. Ensure you run NOOA agents in a sandboxed environment isolated from your primary filesystem, such as NVIDIA OpenShell https://github.com/NVIDIA/OpenShell . NOOA validates generated code AST checks and applies module deny-lists before execution. These are defense-in-depth guardrails, not a containment boundary. They exist to keep generated code from freezing the event loop and to catch common mistakes early — not to stop code that is actively trying to escape. A static checker over Python cannot provide that guarantee: open gives arbitrary file access, importlib can load modules straight from a path, and reflection reaches the rest. The containment boundary is OS-level isolation — always run agents that execute generated code inside a sandbox such as a container, VM, or NVIDIA OpenShell https://github.com/NVIDIA/OpenShell . Do not rely on the in-process validators alone. Choose from supported hosted or local LiteLLM-supported https://docs.litellm.ai/ model: python from nooa.unifiedllm.registry import get llm client llm = get llm client "claude-haiku-4-5" Anthropic after export ANTHROPIC API KEY=... llm = get llm client "gpt-5-mini" OpenAI after export OPENAI API KEY=... llm = get llm client "ollama chat/qwen3:1.7b", api base="http://localhost:11434" Ollama no key llm = get llm client "hosted vllm/Qwen/Qwen3-1.7B", api base="http://localhost:8000/v1" vLLM no key Agents are Python objects . Methods with ... bodies are generation methods — implemented at runtime by an LLM-driven strategy. The signature defines the contract; the docstring is the prompt. python import asyncio from nooa import Agent class FeedbackAgent Agent, llm=llm : """You are an agent specializing in analyzing customer feedback.""" async def analyze feedback self, text: str - str: """Analyze customer feedback for sentiment and key topics in one sentence.""" ... async def main : agent = FeedbackAgent result = await agent.analyze feedback "Great product, but shipping was slow" print result asyncio.run main Run the same code from your own project with python . You can run the checked-in example: uv run python examples/quickstart/01 first generation method.py Rename analyze feedback to analyze feedback briefly and the output changes — your method name, parameters, and docstring are the prompt. Ready for more? See examples/ /NVIDIA-NeMo/labs-OO-Agents/blob/main/examples/README.md for the full progressive tutorial — structured output, tools, strategies, tracing, context blocks, MCP, and more. Every LLM call, code execution, and method invocation is traced by default — orchestrators, generation methods, and helpers, with parent-child spans preserved. If you installed the CLI and viewer dependencies, start the trace viewer and open the run in your browser: uv run nooa start-dev trace viewer on http://localhost:5001 If the viewer isn't running, tracing is silently disabled — no configuration needed either way. — the full progressive tutorial: structured output, tools via examples/README.md /NVIDIA-NeMo/labs-OO-Agents/blob/main/examples/README.md self , strategies, progressive disclosure with doc , tracing, dynamic prompts, context blocks, summarization, skills, MCP, sandbox, and more.— design principles, harness details, capability tests, and SWE-bench Verified / Terminal-Bench 2.0 results. Paper https://arxiv.org/abs/2607.20709 — conventions used inside this repo helpful when reading the source . AGENTS.md /NVIDIA-NeMo/labs-OO-Agents/blob/main/AGENTS.md For a local editable install, clone the repo and sync the development environment with uv : git clone https://github.com/NVIDIA-NeMo/labs-OO-Agents.git cd labs-OO-Agents uv sync --group dev This installs the core framework, workspace packages, development tools, the nooa CLI, and the trace viewer runtime in the repo's .venv . Run CLI commands through uv : uv run nooa --help uv run nooa start-dev trace viewer on http://localhost:5001 Enable pre-commit hooks and run the test/lint suite: uv run pre-commit install uv run pytest run tests uv run ruff check lint uv run pyright type check See CONTRIBUTING.md /NVIDIA-NeMo/labs-OO-Agents/blob/main/CONTRIBUTING.md for the full workflow. If you use NVIDIA-labs OO Agents in your research, please cite: @techreport{nvidia oo agents 2026, title = {NVIDIA-labs OO Agents: Native Python Object-Oriented Agents}, author = {Furgale, Paul and Klingler, Severin and Nolan, James and Staats, Matt and Di Lorenzo, Gaia and Martinez Abad, Elisa and Schueler, Christian and Dinu, Razvan and Devoto, Alessio and Berard, Pascal and Kaplun, Gal and Sarafian, Elad and Roveri, Riccardo and Derczynski, Leon and Silveira Cabral, Ricardo}, year = {2026}, } Apache 2.0. See LICENSE /NVIDIA-NeMo/labs-OO-Agents/blob/main/LICENSE and THIRD PARTY NOTICES.md /NVIDIA-NeMo/labs-OO-Agents/blob/main/THIRD PARTY NOTICES.md .