### Install Observability Example

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/examples/observability/simple_calculator_observability/README.md

Install the necessary packages for the observability example.

```bash
uv pip install -e examples/observability/simple_calculator_observability
```

--------------------------------

### Install Example Workflow

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/examples/MCP/simple_auth_mcp/README.md

Installs the simple_auth_mcp example workflow.

```bash
uv pip install -e examples/MCP/simple_auth_mcp
```

--------------------------------

### Install Simple Calculator Example

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/docs/source/improve-workflows/sizing-calc.md

Install the simple calculator example to use with the sizing calculator.

```bash
uv pip install -e examples/evaluation_and_profiling/simple_calculator_eval
```

--------------------------------

### Install Example Package

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/examples/front_ends/per_user_workflow/README.md

Install the per-user workflow example package using uv pip.

```bash
uv pip install -e ./examples/front_ends/per_user_workflow
```

--------------------------------

### Install Math Assistant Example

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/examples/A2A/math_assistant_a2a_protected/README.md

Install the math assistant example from its source directory using uv pip.

```bash
uv pip install -e examples/A2A/math_assistant_a2a_protected
```

--------------------------------

### Install Simple Calculator Example

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/docs/source/run-workflows/observe/observe-workflow-with-weave.md

Install the simple_calculator example workflow to demonstrate Weave integration.

```bash
uv pip install -e examples/observability/simple_calculator_observability
```

--------------------------------

### Install Example Dependencies

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/examples/a365_example/README.md

Installs the necessary dependencies for the A365 worker example using uv.

```bash
uv sync --project examples/a365_example
```

--------------------------------

### Install Example

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/docs/source/resources/contributing/index.md

Install a specific NeMo Agent Toolkit example using 'uv pip install -e'. This command installs the example in editable mode.

```bash
uv pip install -e ./examples/getting_started/simple_web_query

```

--------------------------------

### Install Simple Auth Example

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/examples/front_ends/simple_auth/README.md

Installs the simple_auth example using uv pip. Ensure you are in the correct project directory.

```bash
uv pip install -e examples/front_ends/simple_auth
```

--------------------------------

### Install Evaluation Example

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/examples/evaluation_and_profiling/simple_web_query_eval/README.md

Install this evaluation example using pip. Ensure you are in the correct development environment.

```bash
uv pip install -e examples/evaluation_and_profiling/simple_web_query_eval
```

--------------------------------

### Install Simple Calculator Example

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/examples/notebooks/mcp_setup_and_integration.ipynb

Installs the 'simple_calculator' example from the NeMo Agent Toolkit repository. It supports installation from a local path in editable mode or from a specified git branch.

```python
from IPython.core.error import StdinNotImplementedError

# Simple input prompt for branch selection
print("=" * 60)
print("Simple Calculator Installation")
print("=" * 60)
print("\nOptions:")
print("  - Enter 'local' for editable install from local repository")
print("  - Enter a branch name (e.g., 'develop', 'main') for git install")
print("=" * 60)

branch_name: str | None = None
try:
    branch_name = input("\nEnter your choice: ").strip()
except StdinNotImplementedError:
    pass

if branch_name is None or branch_name.lower() == 'local':
    # Local editable install
    print("\nInstalling simple calculator in editable mode from local repository...")

    # Try to find the local path relative to current directory
    from pathlib import Path
    # path-check-skip-next-line
    local_path = Path('../../examples/getting_started/simple_calculator')

    if local_path.exists():
        get_ipython().system(f'pip install -e {local_path}')
        print(f"✓ Installed from local path: {local_path.absolute()}")
    else:
        print(f"✗ Error: Local path not found: {local_path.absolute()}")
        print("Make sure you're running this from the correct directory")
else:
    # Git install from specified branch
    print(f"\nInstalling simple calculator from branch: {branch_name}")
    get_ipython().system(f'pip install --no-deps "git+https://github.com/NVIDIA/NeMo-Agent-Toolkit.git@{branch_name}#subdirectory=examples/getting_started/simple_calculator"')
    print(f"✓ Installed from git branch: {branch_name}")

print("\n" + "=" * 60)
```

--------------------------------

### Install Example Package

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/examples/frameworks/strands_demo/bedrock_agentcore/README.md

Installs the example package for the NeMo Agent Toolkit using pip.

```bash
uv pip install -e examples/frameworks/strands_demo
```

--------------------------------

### Install Example Dependencies

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/examples/prompt_from_file/README.md

Install the example's Python dependencies using uv pip. This command should be run from the repository root.

```bash
# From repository root
uv pip install -e examples/prompt_from_file
```

--------------------------------

### Install Multi-Frameworks Workflow

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/examples/frameworks/multi_frameworks/README.md

Install the multi-frameworks example locally using uv pip. Ensure you have the development environment set up.

```bash
uv pip install -e examples/frameworks/multi_frameworks
```

--------------------------------

### Install Example Package

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/examples/finetuning/rl_with_openpipe_art/README.md

Installs the current example package in editable mode within your NeMo Agent Toolkit environment.

```bash
uv pip install -e examples/finetuning/rl_with_openpipe_art
```

--------------------------------

### Install Workflow Example

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/examples/control_flow/parallel_executor/README.md

Install the parallel_executor example workflow using pip. Ensure you are in the root directory of the NeMo Agent Toolkit repository.

```bash
uv pip install -e examples/control_flow/parallel_executor
```

--------------------------------

### Start A2A Server Example

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/docs/source/reference/cli.md

Example of starting an A2A server with a specified configuration file, agent name, and description. The server will run on default host and port.

```bash
nat a2a serve --config_file examples/getting_started/simple_calculator/configs/config.yml \
              --name "Calculator Agent" \
              --description "A calculator agent for mathematical operations"
```

--------------------------------

### Install NeMo Agent Toolkit Example

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/examples/MCP/service_account_auth_mcp/README.md

Installs the example workflow for MCP service account authentication using pip.

```bash
uv pip install -e examples/MCP/service_account_auth_mcp
```

--------------------------------

### Install NeMo Agent Toolkit Examples

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/examples/agents/auto_memory_wrapper/README.md

Install the example agents from the NeMo Agent Toolkit repository. This command should be run from the root directory of the repository.

```bash
uv pip install -e examples/agents
```

--------------------------------

### Install NeMo Agent Toolkit and Example

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/docs/source/build-workflows/llms/using-local-llms.md

Install the NeMo Agent Toolkit and the simple web query example using pip. Ensure your current working directory is the root of the repository.

```bash
uv pip install -e .
uv pip install -e examples/getting_started/simple_web_query
```

--------------------------------

### Install and Start Phoenix Server using CLI

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/examples/frameworks/auto_wrapper/langchain_deep_research/langgraph_deep_research.ipynb

Installs the Arize Phoenix CLI and starts the server in the background. This is an alternative to using Docker.

```bash
%%bash
# Install Phoenix
uv pip install arize-phoenix

# Start Phoenix server in the background
# This will start the server on http://localhost:6006
nohup phoenix serve > /dev/null 2>&1 &

echo "Phoenix server is starting at http://localhost:6006"
echo "Give it a few seconds to fully initialize"
```

--------------------------------

### Install Phoenix Integration and Start Phoenix

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/examples/experimental/codex_agent_adapter/README.md

Install the Phoenix integration using pip and start the Phoenix server. Ensure the correct version is used.

```bash
uv pip install -e packages/nvidia_nat_phoenix
docker run -it --rm -p 4317:4317 -p 6006:6006 arizephoenix/phoenix:13.22
```

--------------------------------

### Full Installation and Setup Script

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/examples/dynamo_integration/latency_sensitivity_demo/INSTALL_LIBRARY.md

A comprehensive script for setting up a clean Ubuntu machine for Dynamo development. It installs system packages, CUDA, Rust, uv, clones the repository, sets up a virtual environment, builds the project, installs SGLang, and verifies the installation.

```bash
# System packages
sudo apt-get update
sudo apt-get install -y build-essential cmake pkg-config python3-dev \
  libclang-dev protobuf-compiler libhwloc-dev libudev-dev

# CUDA toolkit (needed for GPU inference backends)
sudo apt-get install -y nvidia-cuda-toolkit
sudo ln -sf /usr /usr/local/cuda

# Rust
curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y
source "$HOME/.cargo/env"

# uv
curl -LsSf https://astral.sh/uv/install.sh | sh

# Clone and enter repo
git clone https://github.com/ai-dynamo/dynamo.git
cd dynamo

# Virtual environment
uv venv
source .venv/bin/activate

# Build
uv pip install pip maturin
cd lib/bindings/python && maturin develop --uv && cd ../../..
uv pip install -e lib/gpu_memory_service
uv pip install -e .

# SGLang from source (recommended)
git clone https://github.com/sgl-project/sglang.git
uv pip install -e sglang/python

# Verify
python -c "from dynamo.runtime import DistributedRuntime; print('OK')"
```

--------------------------------

### Install Phoenix and Run Relay

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/examples/experimental/hermes_agent_adapter/README.md

Install the Phoenix integration and start the Phoenix server. Then, run the Relay/Phoenix configuration with a specified input.

```bash
uv pip install -e packages/nvidia_nat_phoenix
docker run -it --rm -p 4317:4317 -p 6006:6006 arizephoenix/phoenix:13.22
```

```bash
nat run \
  --config_file examples/experimental/hermes_agent_adapter/configs/config-relay-phoenix.yml \
  --input "Read exactly examples/experimental/hermes_agent_adapter/README.md and summarize the Run With NeMo Relay section. Do not edit files."
```

--------------------------------

### Run Simple Calculator Observability Example

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/docs/source/run-workflows/observe/observe-workflow-with-arize-ax.md

Execute the simple calculator observability example with Arize AX configuration. This command installs the example's dependencies and runs the NeMo Agent Toolkit with the specified configuration file and input.

```bash
uv pip install -e examples/observability/simple_calculator_observability/
nat run --config_file examples/observability/simple_calculator_observability/configs/config-arize-ax.yml --input "What is 2 * 4?"
```

--------------------------------

### Install Tic Tac Toe Example

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/examples/finetuning/dpo_tic_tac_toe/README.md

Installs the DPO Tic Tac Toe example from source using pip. Ensure you have followed the NeMo Agent Toolkit Installation Guide beforehand.

```bash
uv pip install -e examples/finetuning/dpo_tic_tac_toe
```

--------------------------------

### Run Hello World Example

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/README.md

Run the Hello World example using the 'nat' CLI with the specified configuration file and input.

```bash
nat run --config_file workflow.yml --input "List five subspecies of Aardvarks"
```

--------------------------------

### Set Up Sizing Calculator Configuration

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/docs/source/improve-workflows/sizing-calc.md

Set the output directory and configuration file path, create the directory, and copy the example configuration file.

```bash
export CALC_OUTPUT_DIR=.tmp/sizing_calc/
export CONFIG_FILE=${CALC_OUTPUT_DIR}config-sizing-calc.yml

mkdir -p ${CALC_OUTPUT_DIR}

cp examples/evaluation_and_profiling/simple_calculator_eval/configs/config-sizing-calc.yml $CONFIG_FILE
```

--------------------------------

### Run Example with FastAPI

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/examples/prompt_from_file/README.md

Start the agent as a FastAPI HTTP server using the specified configuration file.

```bash
# Run with FastAPI (HTTP server)
nat start fastapi --config_file examples/prompt_from_file/configs/config.yml
```

--------------------------------

### Build and Preview Documentation

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/docs/README.md

Builds the documentation using the make command and then starts a local HTTP server to preview the built HTML files. The output is generated in `docs/build/html`.

```bash
make -C docs
```

```bash
python -m http.server --directory docs/build/html 8000
```

--------------------------------

### Install Workflow with uv pip

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/examples/MCP/simple_calculator_fastmcp/README.md

Install this specific workflow example using uv pip. This ensures all necessary dependencies for the example are available.

```bash
uv pip install -e examples/MCP/simple_calculator_fastmcp
```

--------------------------------

### Install Example Package and Run Profiler

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/examples/dynamo_integration/latency_sensitivity_demo/README.md

Install the latency sensitivity demo package and set the NVIDIA API key. Then, run the 'nat eval' command to profile the workflow and build the prediction trie.

```bash
# Install the example package
uv pip install -e ./examples/dynamo_integration/latency_sensitivity_demo

# Set your NVIDIA API key
export NVIDIA_API_KEY=nvapi-...

# Run profiling (8*8 queries, ~30 seconds each rep)
nat eval --reps 2 --config_file examples/dynamo_integration/latency_sensitivity_demo/src/latency_sensitivity_demo/configs/config_profile.yml
```

--------------------------------

### Run Example with Console

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/examples/prompt_from_file/README.md

Start the agent in interactive console mode, providing the configuration file and an initial input message.

```bash
# Run with console (interactive)
nat start console --config_file examples/prompt_from_file/configs/config.yml --input "What is 5 + 3?"
```

--------------------------------

### Install NVIDIA nat Package and Example

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/examples/frameworks/strands_demo/bedrock_agentcore/README.md

Installs the nvidia-nat package and the example package using uv. Ensures NAT_VERSION build-arg is set and installs dependencies with specific link modes and bytecode compilation. Cleans up __pycache__ and .pyc/.pyo files.

```dockerfile
RUN --mount=type=cache,id=uv_cache,target=/root/.cache/uv,sharing=locked \
    test -n "${NAT_VERSION}" || { echo "NAT_VERSION build-arg is required" >&2; exit 1; } && \
    export SETUPTOOLS_SCM_PRETEND_VERSION=${NAT_VERSION} && \
    export SETUPTOOLS_SCM_PRETEND_VERSION_NVIDIA_NAT=${NAT_VERSION} && \
    export SETUPTOOLS_SCM_PRETEND_VERSION_NVIDIA_NAT_LANGCHAIN=${NAT_VERSION} && \
    export SETUPTOOLS_SCM_PRETEND_VERSION_NVIDIA_NAT_TEST=${NAT_VERSION} && \
    export SETUPTOOLS_SCM_PRETEND_VERSION_FOR_NAT_SIMPLE_CALCULATOR=${NAT_VERSION} && \
    export SETUPTOOLS_SCM_PRETEND_VERSION_FOR_NAT_STRANDS_DEMO=${NAT_VERSION} && \
    uv venv --python ${PYTHON_VERSION} /workspace/.venv && \
    uv sync --link-mode=copy --compile-bytecode --python ${PYTHON_VERSION} && \
    uv pip install -e '.[opentelemetry,phoenix,weave]' --link-mode=copy --compile-bytecode --python ${PYTHON_VERSION} && \
    uv pip install -e ./examples/frameworks/strands_demo --link-mode=copy && \
    uv pip install boto3 aws-opentelemetry-distro && \
    find /workspace/.venv -type d -name __pycache__ -exec rm -rf {} + 2>/dev/null || true && \
    find /workspace/.venv -type f -name "*.pyc" -delete && \
    find /workspace/.venv -type f -name "*.pyo" -delete && \
    find /workspace/.venv -name "*.dist-info" -type d -exec rm -rf {}/RECORD {} + 2>/dev/null || true && \
    rm -rf /workspace/.venv/lib/python*/site-packages/pip /workspace/.venv/lib/python*/site-packages/setuptools
```

--------------------------------

### Install mcp-server-time Dependency

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/examples/MCP/simple_calculator_fastmcp/README.md

Manually install the 'mcp-server-time' package if it was not included when installing the example workflow. This is required for the client to function correctly.

```bash
uv pip install mcp-server-time
```

--------------------------------

### Start Math Assistant Server

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/examples/A2A/math_assistant_a2a/README.md

Launches the math assistant as a FastAPI server using a specified configuration file.

```bash
# Start the math assistant as a FastAPI server on terminal 2
nat serve --config_file examples/A2A/math_assistant_a2a/configs/config.yml
```

--------------------------------

### Start Dynamo and Monitoring

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/external/dynamo/monitoring/README.md

Start the Dynamo stack, which includes the monitoring setup automatically. Alternatively, navigate to the monitoring directory and start Docker Compose manually.

```bash
bash start_dynamo_optimized_thompson_hints_vllm.sh
```

```bash
cd monitoring
docker compose up -d
```

--------------------------------

### Install Cursor Agent Adapter

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/examples/experimental/cursor_agent_adapter/README.md

Install the workflow package using uv pip. Ensure you have followed the general NeMo Agent Toolkit installation guide first.

```bash
uv pip install -e examples/experimental/cursor_agent_adapter
```

--------------------------------

### Install and Serve Simple Calculator Workflow

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/docs/source/reference/rest-api/api-server-endpoints.md

Install the simple calculator workflow and start the NeMo Agent Toolkit server. Ensure the workflow is running on http://localhost:8000 before proceeding.

```bash
uv pip install -e examples/getting_started/simple_calculator
nat serve --config_file examples/getting_started/simple_calculator/configs/config.yml
```

--------------------------------

### Start Phoenix Observability Tool

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/examples/memory/redis/README.md

Start the Phoenix observability tool on localhost:6006 using Docker Compose. This tool is configured for the examples.

```bash
docker compose -f examples/deploy/docker-compose.phoenix.yml up
```

--------------------------------

### Install QEMU for Multi-Architecture Emulation

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/docs/source/resources/contributing/testing/running-ci-locally.md

On apt-based systems, install QEMU utilities required for emulating different architectures. This is a prerequisite for running CI on non-native architectures.

```bash
sudo apt install qemu-utils qemu-system-arm qemu-user-static
```

--------------------------------

### Install and Run Harbor

Source: https://github.com/nvidia/nemo-agent-toolkit/blob/develop/examples/notebooks/eval_harbor_atif_interop.ipynb

Installs Harbor using 'uv tool install' and then runs Harbor with specified dataset, agent, model, and task count. Captures and prints stdout/stderr.

```python
import json
from pathlib import Path

subprocess.run(
    ["uv", "tool", "install", "harbor"],
    check=True,
    capture_output=True,
)
print("Harbor installed.\n")

env = {
    **os.environ,
    "NVIDIA_NIM_API_KEY": os.environ["NVIDIA_API_KEY"],
    "MSWEA_COST_TRACKING": "ignore_errors",
}
print(
    f"Running: harbor run --dataset {HARBOR_DATASET} "
    f"--agent {HARBOR_AGENT} --n-tasks {HARBOR_N_TASKS}\n"
)
result = subprocess.run(
    [
        "harbor", "run",
        "--dataset", HARBOR_DATASET,
        "--agent", HARBOR_AGENT,
        "--model", HARBOR_MODEL,
        "--n-tasks", str(HARBOR_N_TASKS),
        "--agent-env", "MSWEA_COST_TRACKING=ignore_errors",
    ],
    env=env,
    capture_output=True,
    text=True,
    check=False,
)
stdout = result.stdout[-1000:] if len(result.stdout) > 1000 else result.stdout
print(stdout)
if result.returncode != 0:
    print(f"Harbor exited with code {result.returncode}")
    if result.stderr:
        print(result.stderr[-500:])
```