Skip to content

SAMPLE_GUIDELINES.md

Latest commit

 

History

History
200 lines (134 loc) · 9.11 KB

SAMPLE_GUIDELINES.md

File metadata and controls

200 lines (134 loc) · 9.11 KB
Raw

Sample Guidelines

Samples are extremely important for developers to get started with Agent Framework. We strive to provide a wide range of samples that demonstrate the capabilities of Agent Framework with consistency and quality. This document outlines the guidelines for creating samples.

File Structure

Every sample file should follow this order:

  1. Copyright header: # Copyright (c) Microsoft. All rights reserved.
  2. PEP 723 inline script metadata (if external dependencies are needed)
  3. Module docstring: """This sample demonstrates..."""
  4. from __future__ import annotations when applicable
  5. Required imports (including from dotenv import load_dotenv)
  6. Environment variable loading: load_dotenv()
  7. Helper functions
  8. Main function(s) demonstrating functionality
  9. Entry point: use if __name__ == "__main__": asyncio.run(main()) for async samples or if __name__ == "__main__": main() for synchronous samples

When modifying samples, update associated README files in the same or parent folders.

External Dependencies

When samples depend on external packages not included in the dev environment (e.g., semantic-kernel, autogen-agentchat, pandas), declare them using PEP 723 inline script metadata immediately after the copyright header and before the module docstring:

# Copyright (c) Microsoft. All rights reserved.

# /// script
# requires-python = ">=3.10"
# dependencies = [
#     "some-external-package",
# ]
# ///
# Run with any PEP 723 compatible runner, e.g.:
#   uv run samples/path/to/script.py

"""Demonstrate a feature that requires a sample-only dependency."""

This makes samples self-contained and runnable without installing extra packages into the dev environment. Do not add sample-only dependencies to the root pyproject.toml dev group.

Azure Credentials

Always use AzureCliCredential in samples — never DefaultAzureCredential. AzureCliCredential is explicit about the authentication method (requires az login) and avoids unexpected credential resolution that can confuse newcomers.

from azure.identity import AzureCliCredential

credential = AzureCliCredential()

Environment Variables

Basic / Getting Started Samples

For getting started samples (01-get-started/) and basic samples, use explicit placeholder values for non-sensitive parameters to make the code immediately readable:

# Copyright (c) Microsoft. All rights reserved.

"""Demonstrate a basic Foundry-backed agent."""

import asyncio

from agent_framework import Agent
from agent_framework.foundry import FoundryChatClient
from azure.identity import AzureCliCredential


async def main() -> None:
    client = FoundryChatClient(
        project_endpoint="https://your-account.services.ai.azure.com/api/projects/your-project",
        model="gpt-4o",
        credential=AzureCliCredential(),
    )
    agent = Agent(client=client, name="MyAgent", instructions="You are helpful.")
    result = await agent.run("Hello!")
    print(result)

Basic samples should NOT use os.environ, load_dotenv(), or .env files. The placeholder values make the code self-documenting.

Advanced Samples

For advanced samples that require real credentials or multiple configuration values, use environment variables with os.environ and document the required variables in the module docstring:

# Copyright (c) Microsoft. All rights reserved.

"""Demonstrate advanced feature X.

Environment variables:
    FOUNDRY_PROJECT_ENDPOINT - Microsoft Foundry project endpoint.
    FOUNDRY_MODEL - Model deployment name, for example `gpt-4o`.
"""

import asyncio
import os

from agent_framework import Agent
from agent_framework.foundry import FoundryChatClient
from azure.identity import AzureCliCredential

When a change adds, removes, or renames a sample-facing environment variable in repo-level samples or package-local sample documentation for a package included by agent-framework-core[all], update the consolidated inventory in samples/README.md in the same change.

Default Client for Samples

Unless a sample is specifically demonstrating a particular provider (OpenAI direct, Anthropic, Ollama, etc.), use FoundryChatClient from agent_framework.azure as the default client. This is the recommended client for Microsoft Foundry deployments.

Provider-specific samples belong in 02-agents/providers/<provider>/ and should use the provider's native client (e.g., OpenAIChatClient for OpenAI, AnthropicClient for Anthropic).

Syntax Checking

Run uv run poe pyright -S to check samples for syntax errors and missing imports from agent_framework. This uses a relaxed pyright configuration that validates imports without strict type checking.

Some samples depend on external packages (e.g., azure.ai.agentserver.agentframework, microsoft_agents) that are not installed in the dev environment. These are excluded in pyrightconfig.samples.json. When adding or modifying these excluded samples, add them to the exclude list and manually verify they have no import errors from agent_framework packages by temporarily removing them from the exclude list and running the check.

General Guidelines

  • Clear and Concise: Samples should be clear and concise. They should demonstrate a specific set of features or capabilities of Agent Framework. The less concepts a sample demonstrates, the better.
  • Consistent Structure: All samples should have a consistent structure. This includes the folder structure, file naming, and the content of the sample.
  • Incremental Complexity: Samples should start simple and gradually increase in complexity. This helps developers understand the concepts and features of Agent Framework.
  • Documentation: Samples should be over-documented.

Clear and Concise

Try not to include too many concepts in a single sample. The goal is to demonstrate a specific feature or capability of Agent Framework. If you find yourself including too many concepts, consider breaking the sample into multiple samples. A good example of this is to break non-streaming and streaming modes into separate samples.

Consistent Structure

! TODO: Update folder structure to our new needs. ! TODO: Decide on single samples folder or also samples in extensions

Getting Started Samples

The getting started samples are the simplest samples that require minimal setup. These samples should be named in the following format: step<number>_<name>.py. One exception to this rule is when the sample is a notebook, in which case the sample should be named in the following format: <number>_<name>.ipynb.

Incremental Complexity

Try to do a best effort to make sure that the samples are incremental in complexity. For example, in the getting started samples, each step should build on the previous step, and the concept samples should build on the getting started samples, same with the demos.

Documentation

Try to over-document the samples. This includes comments in the code, README.md files, and any other documentation that is necessary to understand the sample. We use the guidance from PEP8 for comments in the code, with a deviation for the initial summary comment in samples and the output of the samples.

For the getting started samples and the concept samples, we should have the following:

  1. A README.md file is included in each set of samples that explains the purpose of the samples and the setup required to run them.

  2. The module docstring should appear after the copyright header and optional metadata comments, before any future imports or other imports. It should explain the purpose of the sample and the required components or concepts. For example:

    # Copyright (c) Microsoft. All rights reserved.

"""Demonstrate a chatbot built from an agent and a chat client.

The agent manages instructions and conversation state. The chat client
generates responses to user messages.
"""

from __future__ import annotations

import asyncio

  3. Mark the code with comments to explain the purpose of each section of the code. For example:

    # 1. Create the instance of the Kernel to register the plugin and service.
...

# 2. Create the agent with the kernel instance.
...

    This will also allow the sample creator to track if the sample is getting too complex.

  4. At the end of the sample, include a section that explains the expected output of the sample. For example:

    """
Sample output:
User:> Why is the sky blue in one sentence?
Mosscap:> The sky is blue due to the scattering of sunlight by the molecules in the Earth's atmosphere,
a phenomenon known as Rayleigh scattering, which causes shorter blue wavelengths to become more
prominent in our visual perception.
"""

For the demos, a README.md file must be included that explains the purpose of the demo and how to run it. The README.md file should include the following:

  • A description of the demo.
  • A list of dependencies required to run the demo.
  • Instructions on how to run the demo.
  • Expected output of the demo.