<!--
SPDX-FileCopyrightText: Copyright (c) 2024-2026, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
SPDX-License-Identifier: Apache-2.0
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
-->
# NVIDIA NeMo Agent Toolkit
> [!WARNING]
> ⚠️ **Deprecation Notice:** NVIDIA NeMo Agent Toolkit has been deprecated as of October 2026 and will reach end of life (EOL) by end of year **2026**. Depending on your needs, migrate to: [NeMo Relay](<https://github.com/NVIDIA/NeMo-Relay>) for observability, middleware, and governance. [NeMo Helix](<https://github.com/NVIDIA-NeMo/nemo-helix>) for a unified CLI with evaluation, optimization, and hardening capabilities. [NeMo Fabric](<https://github.com/NVIDIA/NeMo-Fabric>) for a consistent interface to run agent harnesses and custom agents.
> We recommend completing your migration before the EOL date.
<!-- vale off (due to hyperlinks) -->
[](https://opensource.org/licenses/Apache-2.0)
[](https://github.com/NVIDIA/NeMo-Agent-Toolkit/releases)
[](https://pypi.org/project/nvidia-nat/)
[](https://github.com/NVIDIA/NeMo-Agent-Toolkit/issues)
[](https://github.com/NVIDIA/NeMo-Agent-Toolkit/pulls)
[](https://github.com/NVIDIA/NeMo-Agent-Toolkit)
[](https://github.com/NVIDIA/NeMo-Agent-Toolkit/network/members)
[](https://deepwiki.com/NVIDIA/NeMo-Agent-Toolkit)
[](https://colab.research.google.com/github/NVIDIA/NeMo-Agent-Toolkit/)
<!-- vale on -->
<div align="center">
*NVIDIA NeMo Agent Toolkit adds intelligence to AI agents across any framework—enhancing speed, accuracy, and decision-making through enterprise-grade instrumentation, observability, and continuous learning.*
</div>
## 🔥 New Features
- [**AI Coding Agent Skills:**](./AGENTS.md) Use focused NeMo Agent Toolkit skills to give coding agents task-specific guidance for building, evaluating, optimizing, and observing workflows.
- [**Dynamo Runtime Intelligence:**](./examples/dynamo_integration/latency_sensitivity_demo/README.md) _(experimental)_ Automatically infer per-request latency sensitivity from agent profiles and apply runtime hints for cache control, load-aware routing, and priority-aware serving.
- [**Agent Performance Primitives (APP):**](https://docs.langchain.com/oss/python/integrations/providers/nvidia#install-2) Introduce framework-agnostic performance primitives that accelerate graph-based agent frameworks such as LangChain, CrewAI, and Agno with parallel execution, speculative branching, and node-level priority routing.
- [**Public Plugin API for Third-Party Tools:**](./docs/source/extend/third-party-plugins.md) Build provider-managed integrations against the public NeMo Agent Toolkit plugin API instead of framework-specific internal packages. Thank you to the maintainers of externally managed plugins including [NeMo-Agent-Toolkit-Tavily](https://github.com/tavily-ai/NeMo-Agent-Toolkit-tavily), [NeMo-Agent-Toolkit-Redis](https://github.com/redis-developer/nemo-agent-toolkit-redis), [NeMo-Agent-Toolkit-ATR](https://github.com/Agent-Threat-Rule/NeMo-Agent-Toolkit-atr), [NeMo-Agent-Toolkit-Memanto](https://github.com/moorcheh-ai/NeMo-Agent-Toolkit-memanto), [AffixIO NeMo Plugin](https://github.com/AffixIO/affixio-nemo-plugin), and [NeMo-Agent-Toolkit-hol-guard](https://github.com/hashgraph-online/NeMo-Agent-Toolkit-hol-guard). This new workflow is intended to help third-party package maintainers bring their integrations into NeMo Agent Toolkit workflows more quickly. Note that these third-party packages are managed outside the NeMo Agent Toolkit repository and may release, change, or run CI on schedules independent of NeMo Agent Toolkit.
- [**LangSmith Native Integration:**](./docs/source/run-workflows/observe/observe-workflow-with-langsmith.md) Observe end-to-end agent execution with native LangSmith tracing, run evaluation experiments, compare outcomes, and manage prompt versions across development and production workflows.
- [**FastMCP Workflow Publishing:**](./docs/source/run-workflows/fastmcp-server.md) Publish NeMo Agent Toolkit workflows as MCP servers using the FastMCP server runtime to simplify MCP-native deployment and integration.
- **Migration notice:** `1.5.0` simplifies package installation and dependency management. See the [Migration Guide](./docs/source/resources/migration-guide.md#v150).
## ✨ Key Features
- 🛠️ **Building Agents**: Accelerate your agent development with tools that make it easier to get your agent into production.
- 🧩 [**Framework Agnostic:**](./docs/source/components/integrations/frameworks.md) Work side-by-side with agentic frameworks to add the instrumentation necessary for observing, profiling, and optimizing your agents. Use the toolkit with popular frameworks such as [LangChain](https://www.langchain.com/), [LlamaIndex](https://www.llamaindex.ai/), [CrewAI](https://www.crewai.com/), [Microsoft Semantic Kernel](https://learn.microsoft.com/en-us/semantic-kernel/), and [Google ADK](https://google.github.io/adk-docs/), as well as custom enterprise agentic frameworks and simple Python agents.
- 🔁 [**Reusability:**](./docs/source/components/sharing-components.md) Build components once and use them multiple times to maximize the value from development effort.
- ⚡ [**Customization:**](docs/source/get-started/tutorials/customize-a-workflow.md) Start with a pre-built agent, tool, or workflow, and customize it to your needs.
- 💬 [**Built-In User Interface:**](./docs/source/run-workflows/launching-ui.md) Use the NeMo Agent Toolkit UI chat interface to interact with your agents, visualize output, and debug workflows.
- 📈 **Agent Insights:** Utilize NeMo Agent Toolkit instrumentation to better understand how your agents function at runtime.
- 📊 [**Profiling:**](./docs/source/improve-workflows/profiler.md) Profile entire workflows from the agent level all the way down to individual tokens to identify bottlenecks, analyze token efficiency, and guide developers in optimizing their agents.
- 🔎 [**Observability:**](./docs/source/run-workflows/observe/observe.md) Track performance, trace execution flows, and gain insights into your agent behaviors in production.
- 🚀 **Agent Optimization:** Improve your agent's quality, accuracy, and performance with a suite of tools for all phases of the agent lifecycle.
- 🧪 [**Evaluation System:**](./docs/source/improve-workflows/evaluate.md) Validate and maintain accuracy of agentic workflows with a suite of tools for offline evaluation.
- 🎯 [**Hyper-Parameter and Prompt Optimizer:**](./docs/source/improve-workflows/optimizer.md) Automatically identify the best configuration and prompts to ensure you are getting the most out of your agent.
- 🧠 [**Fine-tuning with Reinforcement Learning:**](./docs/source/improve-workflows/finetuning/index.md) Fine-tune LLMs specifically for your agent and train intrinsic information about your workflow directly into the model.
- ⚡ [**NVIDIA Dynamo Integration:**](./examples/dynamo_integration/README.md) _(experimental)_ Use Dynamo and NeMo Agent Toolkit together to improve agent performance at scale.
- ⚙️ [**Agent Performance Primitives (APP):**](https://docs.langchain.com/oss/python/integrations/providers/nvidia#install-2) Accelerate graph-based agent frameworks such as LangChain, CrewAI, and Agno with parallel execution, speculative branching, and node-level priority routing.
- 🔌 **Protocol Support:** Integrate with common protocols used to build agents.
- 🔗 [**Model Context Protocol (MCP):**](./docs/source/build-workflows/mcp-client.md) Integrate [MCP tools](./docs/source/build-workflows/mcp-client.md) into your agents or serve your tools and agents as an [MCP server](./docs/source/run-workflows/mcp-server.md) for others to consume.
- 🤝 [**Agent-to-Agent (A2A) Protocol:**](./docs/source/components/integrations/a2a.md) Build teams of distributed agents with full support for authentication.
With NeMo Agent Toolkit, you can move quickly, experiment freely, and ensure reliability across all your agent-driven projects.
## 🚀 Installation
Before you begin using NeMo Agent Toolkit, ensure that you have Python 3.11, 3.12, or 3.13 installed on your system.
> [!NOTE]
> For users who want to run the examples, it's required to clone the repository and install from source to get the necessary files required to run the examples. Please refer to the [Examples](./examples/README.md) documentation for more information.
To install the latest stable version of NeMo Agent Toolkit from PyPI, run the following command:
```bash
pip install nvidia-nat
```
NeMo Agent Toolkit has many optional dependencies that can be installed with the core package. Optional dependencies are grouped by framework. For example, to install the LangChain/LangGraph plugin, run the following:
```bash
pip install "nvidia-nat[langchain]"
```
Detailed installation instructions, including the full list of optional dependencies and their conflicts, can be found in the [Installation Guide](./docs/source/get-started/installation.md).
## 🌟 Hello World Example
Before getting started, it's possible to run this simple workflow and many other examples in Google Colab with no setup. Click here to open the introduction notebook: [](https://colab.research.google.com/github/NVIDIA/NeMo-Agent-Toolkit/).
1. Ensure you have set the `NVIDIA_API_KEY` environment variable to allow the example to use NVIDIA NIMs. An API key can be obtained by visiting [`build.nvidia.com`](https://build.nvidia.com/) and creating an account.
```bash
export NVIDIA_API_KEY=<your_api_key>
```
2. Create the NeMo Agent Toolkit workflow configuration file. This file will define the agents, tools, and workflows that will be used in the example. Save the following as `workflow.yml`:
```yaml
functions:
# Add a tool to search wikipedia
wikipedia_search:
_type: wiki_search
max_results: 2
llms:
# Tell NeMo Agent Toolkit which LLM to use for the agent
nim_llm:
_type: nim
model_name: nvidia/nemotron-3.5-lightning-30b-a3b
temperature: 0.0
chat_template_kwargs:
enable_thinking: false
workflow:
# Use an agent that 'reasons' and 'acts'
_type: react_agent
# Give it access to our wikipedia search tool
tool_names: [wikipedia_search]
# Tell it which LLM to use
llm_name: nim_llm
# Make it verbose
verbose: true
# Retry up to 3 times
parse_agent_response_max_retries: 3
```
3. Run the Hello World example using the `nat` CLI and the `workflow.yml` file.
```bash
nat run --config_file workflow.yml --input "List five subspecies of Aardvarks"
```
This will run the workflow and output the results to the console.
```console
Workflow Result:
['Here are five subspecies of Aardvarks:\n\n1. Orycteropus afer afer (Southern aardvark)\n2. O. a. adametzi Grote, 1921 (Western aardvark)\n3. O. a. aethiopicus Sundevall, 1843\n4. O. a. angolensis Zukowsky & Haltenorth, 1957\n5. O. a. erikssoni Lönnberg, 1906']
```
## 📚 Additional Resources
* 📖 [Documentation](https://docs.nvidia.com/nemo/agent-toolkit/latest/): Explore the full documentation for NeMo Agent Toolkit.
* 🧭 [Get Started Guide](./docs/source/get-started/installation.md): Set up your environment and start building with NeMo Agent Toolkit.
* 🤝 [Contributing](./docs/source/resources/contributing/index.md): Learn how to contribute to NeMo Agent Toolkit and set up your development environment.
* 🧪 [Examples](./examples/README.md): Explore examples of NeMo Agent Toolkit workflows located in the [`examples`](./examples) directory of the source repository.
* 🛠️ [Create and Customize NeMo Agent Toolkit Workflows](docs/source/get-started/tutorials/customize-a-workflow.md): Learn how to create and customize NeMo Agent Toolkit workflows.
* 🤖 [AI Coding Agent Skill](./docs/source/resources/contributing/agent-skills.md): Install the NeMo Agent Toolkit skill and use example prompts for agent-assisted development.
* 🎯 [Evaluate with NeMo Agent Toolkit](./docs/source/improve-workflows/evaluate.md): Learn how to evaluate your NeMo Agent Toolkit workflows.
* 🆘 [Troubleshooting](./docs/source/resources/troubleshooting.md): Get help with common issues.
## 🛣️ Roadmap
- [x] Automatic Reinforcement Learning (RL) to fine-tune LLMs for a specific agent.
- [x] Integration with [NVIDIA Dynamo](https://github.com/ai-dynamo/dynamo) to reduce LLM latency at scale.
- [x] Improve agent throughput with KV-Cache optimization.
- [ ] Improved, standalone evaluation harness and migration to [ATIF](https://github.com/harbor-framework/harbor/blob/main/rfcs/0001-trajectory-format.md) for trajectory format.
- [ ] Support for additional programming languages (TypeScript, Rust, Go, WASM) with compiled libraries.
- [ ] Phasing out wrapping architecture to ease onboarding for more agents.
- [ ] Support for adding skills and sandboxes to existing agents.
- [ ] MCP authentication improvements.
- [ ] Improved memory interface to support self-improving agents.
## 📊 Telemetry
The NeMo Agent Toolkit includes runtime telemetry hooks for the `nat` command-line tool to help guide improvements. Telemetry is best-effort and never blocks or fails a CLI invocation. Once you opt in (see below), events are sent to the shared NeMo Usage Telemetry ingest.
### How consent works
The first time you run an interactive `nat` command, you'll see a one-time consent prompt explaining what is collected and asking whether to allow it. The prompt defaults to **yes** (pressing Enter accepts); type `n` to opt out. Your decision is persisted to `~/.config/nat/telemetry.toml` and respected on every subsequent invocation.
In **non-interactive contexts** (CI, cron, piped scripts, daemons), telemetry is **always off** unless you explicitly enable it via the environment variable below. We never send data when there's no opportunity to ask.
You can change your decision anytime:
```bash
nat configure telemetry --enable # opt in
nat configure telemetry --disable # opt out
nat configure telemetry --status # show the current effective state
```
Or override the persisted decision (and skip the prompt) via environment variable:
```bash
export NAT_TELEMETRY_ENABLED=false # disable for this shell session
export NAT_TELEMETRY_ENABLED=true # enable for this shell session
```
The environment variable takes precedence over the persisted file. If both disagree, `nat configure telemetry --status` will tell you which one is winning.
### What is collected
For each `nat` command invocation, a single event is sent at exit containing:
- The top-level command name, such as `run`, `serve`, or `evaluate`.
- The second-level command name when applicable, such as `list-components` for `nat info list-components`.
- The outcome: `success`, `failure`, or `interrupted`.
- The wall-clock duration in milliseconds.
- The process exit code.
- The Python class name of the raised exception on failure (the message is not collected).
- The Python runtime version, such as `3.11.7`.
### What is not collected
The following are never collected:
- Command arguments or option values.
- Workflow names, function names, model names, or any contents of configuration files.
- File paths, hostnames, usernames, IP addresses, or any other identifying information.
- The output of any command.
## 💬 Feedback
We would love to hear from you! Please file an issue on [GitHub](https://github.com/NVIDIA/NeMo-Agent-Toolkit/issues) if you have any feedback or feature requests.
## 🤝 Acknowledgements
We would like to thank the following groups for their contribution to the toolkit:
- [Synopsys](https://www.synopsys.com/)
- Google ADK framework support.
- Microsoft AutoGen framework support.
- [W&B Weave Team](https://wandb.ai/site/weave/)
- Contributions to the evaluation and telemetry system.
In addition, we would like to thank the following open source projects that made NeMo Agent Toolkit possible:
- [Agent2Agent (A2A) Protocol](https://github.com/a2aproject/A2A)
- [CrewAI](https://github.com/crewAIInc/crewAI)
- [Dynamo](https://github.com/ai-dynamo/dynamo)
- [FastAPI](https://github.com/tiangolo/fastapi)
- [Google Agent Development Kit (ADK)](https://github.com/google/adk-python)
- [LangChain](https://github.com/langchain-ai/langchain)
- [Llama-Index](https://github.com/run-llama/llama_index)
- [Mem0ai](https://github.com/mem0ai/mem0)
- [Microsoft AutoGen](https://github.com/microsoft/autogen)
- [MinIO](https://github.com/minio/minio)
- [Model Context Protocol (MCP)](https://github.com/modelcontextprotocol/modelcontextprotocol)
- [OpenTelemetry](https://github.com/open-telemetry/opentelemetry-python)
- [Phoenix](https://github.com/arize-ai/phoenix)
- [Redis](https://github.com/redis/redis-py)
- [Semantic Kernel](https://github.com/microsoft/semantic-kernel)
- [Strands](https://github.com/strands-agents/sdk-python)
- [uv](https://github.com/astral-sh/uv)
- [Weave](https://github.com/wandb/weave)