Setting Up the Agent Environment

Set up your LLM provider to run the pattern examples. Azure OpenAI is the default provider. For a free local option, try Ollama.

Prerequisites

• .NET 10.0 SDK or later - Download
• Git - to clone the patterns repository
• An LLM provider - see options below

Choose Your LLM Provider

---

Option 1: Azure OpenAI

Default | Enterprise

Azure OpenAI is the default provider. For enterprise deployments with Azure compliance and security features.

Prerequisites

• Azure subscription
• Azure OpenAI resource with a deployed model
• API key from your Azure OpenAI resource

Set Environment Variables

PowerShell:

$env:AZURE_OPENAI_API_KEY = "your-azure-api-key"
$env:AZURE_OPENAI_ENDPOINT = "https://your-resource.openai.azure.com/"

Bash:

export AZURE_OPENAI_API_KEY=your-azure-api-key
export AZURE_OPENAI_ENDPOINT=https://your-resource.openai.azure.com/

Or create a .env file in the patterns directory:

AZURE_OPENAI_API_KEY = "your-azure-api-key"
AZURE_OPENAI_ENDPOINT = "https://your-resource.openai.azure.com/"

Run a Pattern

# Provider is specified in code, not environment variables
cd patterns/01-prompt-chaining/src/PromptChaining
dotnet run

---

Option 2: Ollama

Free | Local

Ollama lets you run open-source LLMs locally on your machine. No API keys, no costs, works offline.

Step 1: Install Ollama

Download and install from ollama.com

Step 2: Start Ollama

Windows/Mac: Ollama runs automatically after installation. Look for the llama icon in your system tray.

Linux: Start the Ollama server:

ollama serve

Step 3: Pull a Model

ollama pull llama3.2

For smaller machines, try phi3 or llama3.2:1b

Step 4: Set Environment Variables (Optional)

Ollama uses http://localhost:11434 by default. Only set this if using a custom endpoint:

PowerShell:

$env:OLLAMA_ENDPOINT = "http://localhost:11434"

Bash:

export OLLAMA_ENDPOINT=http://localhost:11434

Or create a .env file in the patterns directory:

OLLAMA_ENDPOINT = "http://localhost:11434"  # optional

Step 5: Run a Pattern

cd patterns/01-prompt-chaining/src/PromptChaining
dotnet run

---

Option 3: GitHub Models

Free tier | Cloud

Use your GitHub account to access AI models like GPT-4o, Llama, and more. If you have the GitHub CLI installed, setup is just one command.

Step 1: Install GitHub CLI

If you don’t have it, install from cli.github.com

Step 2: Authenticate with GitHub

gh auth login

Step 3: Get Your Token

# This outputs your GitHub token
gh auth token

Step 4: Set Environment Variables

PowerShell:

$env:GITHUB_TOKEN = (gh auth token)

Bash:

export GITHUB_TOKEN=$(gh auth token)

Or create a .env file in the patterns directory:

GITHUB_TOKEN = "github_pat_..."

Available Models

• gpt-4o - OpenAI GPT-4o
• gpt-4o-mini - Fast and affordable GPT-4o
• Llama-3.3-70B-Instruct - Meta’s Llama 3.3
• Mistral-Large-2411 - Mistral Large

See github.com/marketplace/models for the full list.

Step 5: Run a Pattern

cd patterns/01-prompt-chaining/src/PromptChaining
dotnet run

---

Option 4: OpenAI

Pay-as-you-go

Use OpenAI’s API directly with models like GPT-4o.

Step 1: Get an API Key

Create an account at platform.openai.com and generate an API key.

Step 2: Set Environment Variables

PowerShell:

$env:OPENAI_API_KEY = "sk-..."

Bash:

export OPENAI_API_KEY=sk-...

Or create a .env file in the patterns directory:

OPENAI_API_KEY = "sk-..."

Step 3: Run a Pattern

cd patterns/01-prompt-chaining/src/PromptChaining
dotnet run

---

Option 5: OpenRouter

Pay-as-you-go | Multi-Model

OpenRouter provides access to 100+ models from multiple providers (OpenAI, Anthropic, Google, Meta, Mistral, and more) through a single API. Great for trying different models without managing multiple API keys.

Step 1: Get an API Key

Create an account at openrouter.ai and generate an API key from the Keys page.

Step 2: Set Environment Variables

PowerShell:

$env:OPENROUTER_API_KEY = "sk-or-..."

Bash:

export OPENROUTER_API_KEY=sk-or-...

Or create a .env file in the patterns directory:

OPENROUTER_API_KEY = "sk-or-..."

Popular Models

• openai/gpt-4o-mini - Fast and affordable GPT-4o
• anthropic/claude-sonnet-4 - Anthropic’s Claude Sonnet 4
• google/gemini-2.0-flash - Google’s Gemini 2.0 Flash
• meta-llama/llama-3.3-70b-instruct - Meta’s Llama 3.3

See openrouter.ai/models for the full list.

Step 3: Run a Pattern

cd patterns/01-prompt-chaining/src/PromptChaining
dotnet run

---

Troubleshooting

Azure: Authentication failed

Check that your AZURE_OPENAI_API_KEY and AZURE_OPENAI_ENDPOINT are correct. Get your API key from the Azure portal under your Azure OpenAI resource.

Ollama: “connection refused”

Make sure Ollama is running. On Windows/Mac, check the system tray. On Linux, run ollama serve.

Ollama: Model not found

Pull the model first: ollama pull llama3.2

GitHub Models: 401 Unauthorized

Make sure you’re authenticated with gh auth login and your GITHUB_TOKEN is valid. Run gh auth token to verify.

GitHub Models: Model not found

Check the exact model name at github.com/marketplace/models. Model names are case-sensitive.

OpenAI: 401 Unauthorized

Check that your OPENAI_API_KEY is correct and has available credits.

OpenRouter: 401 Unauthorized

Check that your OPENROUTER_API_KEY is correct. Keys start with sk-or-.

OpenRouter: Model not found

Check the model name format at openrouter.ai/models. Models use provider/model-name format.

.env file not loading

Make sure your .env file is in the patterns directory or a parent directory. The file is loaded automatically when running patterns.

---

Next Steps

Now that your environment is set up, try the first pattern:

Start with Prompt Chaining →