NVIDIA NeMo Fabric

NeMo Fabric gives applications one configurable, observable way to run applications across multiple agent harnesses. It standardizes configuration, lifecycle management, and results without requiring a separate integration for every harness.

NeMo Fabric lets you change harnesses without rebuilding each integration, isolate conflicting runtime dependencies, and manage harness configuration, execution, and observability consistently. Every run returns normalized results, artifacts, and telemetry for downstream systems to consume.

It provides:

• a versioned, typed configuration contract;
• ordinary Python composition for experiment variants;
• adapter integrations for harness-specific launch and control;
• a Python SDK backed by the Rust core;
• normalized run results, artifact manifests, and telemetry references.

Install

NeMo Fabric supports Python 3.11 through 3.14. Use Python 3.11 through 3.13 for Hermes Agent; the Harbor integration requires Python 3.12 or later.

Install the core runtime and Python SDK:

pip install nemo-fabric

To install the NeMo Fabric runtime, adapter, and supported harness in one environment, choose one of the following nemo-fabric harness extras:

pip install "nemo-fabric[claude]"
pip install "nemo-fabric[codex]"
pip install "nemo-fabric[deepagents]"
pip install "nemo-fabric[hermes-agent]"

To install an adapter and its harness without the NeMo Fabric runtime, choose one of the following adapter package harness extras:

pip install "nemo-fabric-adapters-claude[harness]"
pip install "nemo-fabric-adapters-codex[harness]"
pip install "nemo-fabric-adapters-deepagents[harness]"
pip install "nemo-fabric-adapters-hermes[harness]"

Every adapter package also provides an adapter-scoped full extra, which does not install the NeMo Fabric runtime. For Claude and Codex, full installs the same dependencies as harness. For LangChain Deep Agents and Hermes Agent, full also installs the NeMo Relay Python package.

If the environment already manages a compatible harness, choose one of the following bare adapter packages:

pip install nemo-fabric-adapters-claude
pip install nemo-fabric-adapters-codex
pip install nemo-fabric-adapters-deepagents
pip install nemo-fabric-adapters-hermes

The adapter distribution contains only adapter-owned runtime dependencies. It does not install the NeMo Fabric runtime. Select harness or full to install the harness. If the runtime shares an environment with an existing compatible harness, install nemo-fabric and the bare adapter package together.

NeMo Fabric supports running an agent harness in a different virtual environment from the NeMo Fabric runtime. This separation can isolate harnesses that have conflicting dependencies. Use matching NeMo Fabric release versions for the runtime and adapter package unless a different pairing has been explicitly validated.

Integrations

Harbor Integration

pip install "nemo-fabric[harbor]"

Relay Integration

pip install "nemo-fabric[relay]"

This installs a version of the NeMo Relay Python package known to be compatible with the installed version of NeMo Fabric.

The LangChain Deep Agents and Hermes Agent adapter packages also provide relay and full extras for environments that do not install nemo-fabric. Claude and Codex require the nemo-relay CLI instead of the NeMo Relay Python package. They do not provide a relay extra. Refer to the NeMo Relay CLI install guide for instructions on installing the CLI tool.

Core Concepts

• Typed configuration: Construct a complete FabricConfig in Python and use ordinary functions to create experiment variants.
• Adapters: Select harness-specific integrations with harness.adapter_id.
• Artifacts: Receive normalized output, logs, patches, and telemetry references through an ArtifactManifest.

The experimental nemo-fabric CLI is distributed separately from the Python package. It selects complete typed configs from built-in presets and maintained examples. Applications that need a stable integration surface should use the Python SDK.

Learn More

Refer to the NVIDIA NeMo Fabric documentation for installation, configuration, and usage guidance. Source code is available in the NVIDIA NeMo Fabric repository.