docs: add per-example READMEs with .env support

.github/workflows/ci.yml

@@ -31,4 +31,8 @@ jobs:
         uses: abatilo/actions-poetry@7b6d33e44b4f08d7021a1dee3c044e9c253d6439
 
       - name: Install dependencies
-        run: poetry install --all-extras
+        run: |
+          for dir in examples/*/; do
+            echo "Installing $dir"
+            poetry -C "$dir" install
+          done

.gitignore

@@ -1,3 +1,4 @@
 .python-version
 poetry.lock
 __pycache__/
+.env

README.md

@@ -8,106 +8,31 @@ This demo requires Python 3.10 or higher.
 
 ## Build Instructions
 
-This repository includes examples for `OpenAI`, `Bedrock`, and `LangChain` for multi-provider support. Depending on your preferred provider, you may have to take some additional steps.
+This repository includes examples for `OpenAI`, `Bedrock`, `Gemini`, `LangChain`, `LangGraph`, `Judge`, and `Observability`. Depending on your preferred provider, you may have to take some additional steps.
 
 ### General setup
 
 1. [Create an AI Config](https://launchdarkly.com/docs/home/ai-configs/create) using the key specified in each example, or copy the key of existing AI Config in your LaunchDarkly project that you want to evaluate.
-1. Set the environment variable `LAUNCHDARKLY_SDK_KEY` to your LaunchDarkly SDK key and `LAUNCHDARKLY_AI_CONFIG_KEY` to the AI Config key; otherwise, an AI Config of `sample-ai-config` or `sample-ai-agent-config` will be assumed for most examples.
-
-   ```bash
-   export LAUNCHDARKLY_SDK_KEY="1234567890abcdef"
-   export LAUNCHDARKLY_AI_CONFIG_KEY="sample-ai-config"
-   ```
 
 1. Ensure you have [Poetry](https://python-poetry.org/) installed.
 
-### Provider-Specific Setup
-
-#### OpenAI setup (single provider)
-
-1. Install the required dependencies with `poetry install -E openai` or `poetry install --all-extras`.
-1. Set the environment variable `OPENAI_API_KEY` to your OpenAI key.
-1. On the command line, run `poetry run openai-example`.
-
-#### Chat with observability (observability plugin example)
-
-This example demonstrates how to use the LaunchDarkly observability SDK plugin to monitor AI chat operations. For more details, see the [Python SDK observability reference](https://launchdarkly.com/docs/sdk/observability/python).
+1. Create a `.env` file in the repository root with at least your LaunchDarkly SDK key:
 
-1. Install the required dependencies with `poetry install -E observability` or `poetry install --all-extras`.
-1. Set the environment variable for your AI provider (e.g., `OPENAI_API_KEY`), or configure your AI Config to use a different provider.
-1. Optionally, set service identification:
-   ```bash
-   export SERVICE_NAME="my-ai-service"
-   export SERVICE_VERSION="1.0.0"
    ```
-1. On the command line, run `poetry run chat-observability-example`.
-
-The observability plugin automatically captures and sends data to LaunchDarkly:
-- **Observability tab**: SDK operations, flag evaluations, error monitoring, logging, and distributed tracing
-- **AI Config Monitoring tab**: Token usage, duration, success/error rates, and custom metadata for filtering and analysis
-
-View your data in the LaunchDarkly dashboard under **Observability** tabs.
-
-#### Bedrock setup (single provider)
-
-1. Install the required dependencies with `poetry install -E bedrock` or `poetry install --all-extras`.
-1. Ensure the required AWS credentials can be [auto-detected by the `boto3` library](https://boto3.amazonaws.com/v1/documentation/api/latest/guide/credentials.html). Examples might include environment variables, role providers, or shared credential files.
-1. On the command line, run `poetry run bedrock-example`.
-
-#### Gemini setup (single provider)
-
-1. Install the required dependencies with `poetry install -E gemini` or `poetry install --all-extras`.
-1. Set the environment variable `GOOGLE_API_KEY` to your Google API key.
-1. On the command line, run `poetry run gemini-example`.
-
-#### LangChain setup (multiple providers)
-
-This example uses `OpenAI`, `Bedrock`, and `Gemini` LangChain provider packages. You can add additional LangChain providers using the `poetry add` command.
-
-1. Install all dependencies with `poetry install -E langchain` or `poetry install --all-extras`.
-1. Set up API keys for the providers you want to use.
-1. On the command line, run `poetry run langchain-example`
-
-#### LangGraph setup (multiple providers, single agent)
-
-1. Install all dependencies with `poetry install -E langgraph` or `poetry install --all-extras`.
-1. Set up API keys for the providers you want to use.
-1. Optionally set this environment variable to use a different agent config:
-   ```bash
-   export LAUNCHDARKLY_AGENT_CONFIG_KEY="sample-ai-agent-config"
+   LAUNCHDARKLY_SDK_KEY=your-launchdarkly-sdk-key
    ```
-1. On the command line, run `poetry run langgraph-agent-example`.
-
-#### LangGraph setup (multiple providers, multiple agents)
 
-1. Install all dependencies with `poetry install -E langgraph` or `poetry install --all-extras`.
-1. Set up API keys for the providers you want to use.
-1. [Create an AI Config (Agent-based)](https://launchdarkly.com/docs/home/ai-configs/agents) using the keys below. Write a goal for each config and enable it with targeting rules.
-1. Optionally set these environment variables to use different agent configs:
-   ```bash
-   export LAUNCHDARKLY_ANALYZER_CONFIG_KEY="code-review-analyzer"
-   export LAUNCHDARKLY_DOCUMENTATION_CONFIG_KEY="code-review-documentation"
-   ```
-1. On the command line, run `poetry run langgraph-multi-agent-example`.
-
-#### Judge setup (judge evaluation)
-
-These examples demonstrate how to use LaunchDarkly's judge functionality to evaluate AI responses for accuracy, relevance, and other metrics.
-
-1. Install dependencies with `poetry install -E langchain` or `poetry install --all-extras`.
-1. Set up API keys for the provider you want to use (OpenAI, Bedrock, or Gemini).
-1. [Create an AI Config](https://launchdarkly.com/docs/home/ai-configs/create) for chat functionality.
-1. [Create a Judge Config](https://launchdarkly.com/docs/home/ai-configs/judges) for evaluation.
-1. Set the required environment variables:
-   ```bash
-   export LAUNCHDARKLY_SDK_KEY="your-sdk-key"
-   export LAUNCHDARKLY_AI_CONFIG_KEY="sample-ai-config"
-   export LAUNCHDARKLY_AI_JUDGE_KEY="sample-ai-judge-accuracy"
-   ```
-   Note: The default values are `sample-ai-config` for AI Config and `sample-ai-judge-accuracy` for Judge Config if not specified.
+   Each example README describes the full set of environment variables needed. The `.env` file is loaded automatically when running any example.
 
-##### Available judge examples:
+### Examples
 
-- **Chat with automatic judge evaluation** (`poetry run chat-judge-example`): Uses the chat functionality which automatically evaluates responses with any judges defined in the AI config.
-- **Direct judge evaluation** (`poetry run direct-judge-example`): Evaluates specific input/output pairs using a judge configuration directly.
+| Example | Description | README |
+| --- | --- | --- |
+| **OpenAI** | Single provider using OpenAI | [examples/openai](examples/openai/README.md) |
+| **Bedrock** | Single provider using AWS Bedrock | [examples/bedrock](examples/bedrock/README.md) |
+| **Gemini** | Single provider using Google Gemini | [examples/gemini](examples/gemini/README.md) |
+| **LangChain** | Multiple providers via LangChain | [examples/langchain](examples/langchain/README.md) |
+| **LangGraph Agent** | Single agent using LangGraph | [examples/langgraph_agent](examples/langgraph_agent/README.md) |
+| **LangGraph Multi-Agent** | Multiple agents using LangGraph | [examples/langgraph_multi_agent](examples/langgraph_multi_agent/README.md) |
+| **Judge** | Judge evaluation of AI responses | [examples/judge](examples/judge/README.md) |
+| **Chat with Observability** | Observability plugin for AI chat monitoring | [examples/chat_observability](examples/chat_observability/README.md) |

examples/bedrock/README.md

@@ -0,0 +1,43 @@
+# Bedrock Example (Single Provider)
+
+This example demonstrates how to use LaunchDarkly's AI Config with the AWS Bedrock provider.
+
+## Prerequisites
+
+- Python 3.10 or higher
+- [Poetry](https://python-poetry.org/) installed
+- A LaunchDarkly account with an [AI Config](https://launchdarkly.com/docs/home/ai-configs/create) created
+- AWS credentials configured for Bedrock access
+
+## Setup
+
+1. Create a `.env` file in this directory with the following variables:
+
+   ```
+   LAUNCHDARKLY_SDK_KEY=your-launchdarkly-sdk-key
+   LAUNCHDARKLY_AI_CONFIG_KEY=sample-ai-config
+   ```
+
+   > `LAUNCHDARKLY_AI_CONFIG_KEY` defaults to `sample-ai-config` if not set.
+
+2. Ensure your AWS credentials can be [auto-detected by the `boto3` library](https://boto3.amazonaws.com/v1/documentation/api/latest/guide/credentials.html). You can set them in your `.env` file:
+
+   ```
+   AWS_ACCESS_KEY_ID=your-access-key-id
+   AWS_SECRET_ACCESS_KEY=your-secret-access-key
+   AWS_DEFAULT_REGION=us-east-1
+   ```
+
+   Other options include role providers or shared credential files.
+
+3. Install the required dependencies:
+
+   ```bash
+   poetry install
+   ```
+
+## Run
+
+```bash
+poetry run bedrock-example
+```

examples/bedrock_example.py → examples/bedrock/bedrock_example.py

@@ -1,11 +1,14 @@
 import os
+from dotenv import load_dotenv
 import ldclient
 from ldclient import Context
 from ldclient.config import Config
 from ldai import LDAIClient
 import boto3
 
-client = boto3.client("bedrock-runtime", region_name="us-east-1")
+load_dotenv()
+
+client = boto3.client("bedrock-runtime", region_name=os.getenv('AWS_DEFAULT_REGION', 'us-east-1'))
 
 # Set sdk_key to your LaunchDarkly SDK key.
 sdk_key = os.getenv('LAUNCHDARKLY_SDK_KEY')

examples/bedrock/pyproject.toml

@@ -0,0 +1,21 @@
+[tool.poetry]
+name = "hello-python-ai-bedrock"
+version = "0.1.0"
+description = "Hello LaunchDarkly for Python AI - Bedrock"
+authors = ["LaunchDarkly <dev@launchdarkly.com>"]
+license = "Apache-2.0"
+readme = "README.md"
+packages = [{include = "bedrock_example.py"}]
+
+[tool.poetry.scripts]
+bedrock-example = "bedrock_example:main"
+
+[tool.poetry.dependencies]
+python = "^3.10"
+python-dotenv = ">=1.0.0"
+launchdarkly-server-sdk-ai = "^0.17.0"
+boto3 = ">=0.2.0"
+
+[build-system]
+requires = ["poetry-core"]
+build-backend = "poetry.core.masonry.api"

examples/chat_observability/README.md

@@ -0,0 +1,48 @@
+# Chat with Observability (Observability Plugin Example)
+
+This example demonstrates how to use the LaunchDarkly observability SDK plugin to monitor AI chat operations. For more details, see the [Python SDK observability reference](https://launchdarkly.com/docs/sdk/observability/python).
+
+The observability plugin automatically captures and sends data to LaunchDarkly:
+
+- **Observability tab**: SDK operations, flag evaluations, error monitoring, logging, and distributed tracing
+- **AI Config Monitoring tab**: Token usage, duration, success/error rates, and custom metadata for filtering and analysis
+
+View your data in the LaunchDarkly dashboard under **Observability** tabs.
+
+## Prerequisites
+
+- Python 3.10 or higher
+- [Poetry](https://python-poetry.org/) installed
+- A LaunchDarkly account with an [AI Config](https://launchdarkly.com/docs/home/ai-configs/create) created
+- An API key for your AI provider (e.g., OpenAI)
+
+## Setup
+
+1. Create a `.env` file in this directory with the following variables:
+
+   ```
+   LAUNCHDARKLY_SDK_KEY=your-launchdarkly-sdk-key
+   LAUNCHDARKLY_AI_CONFIG_KEY=sample-ai-config
+   OPENAI_API_KEY=your-openai-api-key
+   ```
+
+   > `LAUNCHDARKLY_AI_CONFIG_KEY` defaults to `sample-ai-config` if not set.
+
+   Optionally, set service identification:
+
+   ```
+   SERVICE_NAME=my-ai-service
+   SERVICE_VERSION=1.0.0
+   ```
+
+2. Install the required dependencies:
+
+   ```bash
+   poetry install
+   ```
+
+## Run
+
+```bash
+poetry run chat-observability-example
+```

examples/chat_observability_example.py → examples/chat_observability/chat_observability_example.py

@@ -1,4 +1,5 @@
 import os
+from dotenv import load_dotenv
 import asyncio
 import logging
 import ldclient
@@ -7,6 +8,8 @@
 from ldai import LDAIClient, AICompletionConfigDefault
 from ldobserve import ObservabilityConfig, ObservabilityPlugin
 
+load_dotenv()
+
 logging.getLogger('ldclient').setLevel(logging.WARNING)
 
 # Set sdk_key to your LaunchDarkly SDK key.

examples/chat_observability/pyproject.toml

@@ -0,0 +1,22 @@
+[tool.poetry]
+name = "hello-python-ai-chat-observability"
+version = "0.1.0"
+description = "Hello LaunchDarkly for Python AI - Chat with Observability"
+authors = ["LaunchDarkly <dev@launchdarkly.com>"]
+license = "Apache-2.0"
+readme = "README.md"
+packages = [{include = "chat_observability_example.py"}]
+
+[tool.poetry.scripts]
+chat-observability-example = "chat_observability_example:main"
+
+[tool.poetry.dependencies]
+python = "^3.10"
+python-dotenv = ">=1.0.0"
+launchdarkly-server-sdk-ai = "^0.17.0"
+launchdarkly-observability = ">=0.1.0"
+openai = ">=0.2.0"
+
+[build-system]
+requires = ["poetry-core"]
+build-backend = "poetry.core.masonry.api"

examples/gemini/README.md

@@ -0,0 +1,34 @@
+# Gemini Example (Single Provider)
+
+This example demonstrates how to use LaunchDarkly's AI Config with the Google Gemini provider.
+
+## Prerequisites
+
+- Python 3.10 or higher
+- [Poetry](https://python-poetry.org/) installed
+- A LaunchDarkly account with an [AI Config](https://launchdarkly.com/docs/home/ai-configs/create) created
+- A [Google API key](https://aistudio.google.com/apikey)
+
+## Setup
+
+1. Create a `.env` file in this directory with the following variables:
+
+   ```
+   LAUNCHDARKLY_SDK_KEY=your-launchdarkly-sdk-key
+   LAUNCHDARKLY_AI_CONFIG_KEY=sample-ai-config
+   GOOGLE_API_KEY=your-google-api-key
+   ```
+
+   > `LAUNCHDARKLY_AI_CONFIG_KEY` defaults to `sample-ai-config` if not set.
+
+2. Install the required dependencies:
+
+   ```bash
+   poetry install
+   ```
+
+## Run
+
+```bash
+poetry run gemini-example
+```

examples/gemini_example.py → examples/gemini/gemini_example.py

@@ -1,4 +1,5 @@
 import os
+from dotenv import load_dotenv
 import ldclient
 from ldclient import Context
 from ldclient.config import Config
@@ -8,6 +9,8 @@
 from google.genai import types
 from typing import List, Optional, Tuple
 
+load_dotenv()
+
 # Set sdk_key to your LaunchDarkly SDK key.
 sdk_key = os.getenv('LAUNCHDARKLY_SDK_KEY')
 

examples/gemini/pyproject.toml

@@ -0,0 +1,21 @@
+[tool.poetry]
+name = "hello-python-ai-gemini"
+version = "0.1.0"
+description = "Hello LaunchDarkly for Python AI - Gemini"
+authors = ["LaunchDarkly <dev@launchdarkly.com>"]
+license = "Apache-2.0"
+readme = "README.md"
+packages = [{include = "gemini_example.py"}]
+
+[tool.poetry.scripts]
+gemini-example = "gemini_example:main"
+
+[tool.poetry.dependencies]
+python = "^3.10"
+python-dotenv = ">=1.0.0"
+launchdarkly-server-sdk-ai = "^0.17.0"
+google-genai = "^1.30.0"
+
+[build-system]
+requires = ["poetry-core"]
+build-backend = "poetry.core.masonry.api"