Semantic Kernel Python Flask Starter

The sk-python-flask flask application demonstrates how to execute a semantic function within a Flask backend. The application can also be used as a ChatGPT plugin. If not using the starter as a ChatGPT plugin, you may not need the following files and the app routes that serve them: openapi.yaml, .well-known/ai-plugin.json, logo.png

Prerequisites

Python 3.8 and above
Poetry is used for packaging and dependency management
Semantic Kernel Tools

Configuring the starter

This starter can be configured in two ways:

A .env file in the project which holds api keys and other secrets and configurations
Or with HTTP Headers on each request

Make sure you have an Open AI API Key or Azure Open AI service key

Configure with a .env file

Copy the .env.example file to a new file named .env. Then, copy those keys into the .env file:

OPENAI_API_KEY=""
OPENAI_ORG_ID=""
AZURE_OPENAI_DEPLOYMENT_NAME=""
AZURE_OPENAI_ENDPOINT=""
AZURE_OPENAI_API_KEY=""
GORILLA_CLI_PATH="/path/to/gorilla-cli" # Path to the Gorilla CLI executable

Make sure to replace "/path/to/gorilla-cli" with the actual path to the Gorilla CLI on your system.

Configure with HTTP Headers

On each HTTP request, use these headers:

"x-ms-sk-completion-model" # e.g. text-davinci-003
"x-ms-sk-completion-endpoint" # e.g. https://my-endpoint.openai.azure.com
"x-ms-sk-completion-backend" # AZURE_OPENAI or OPENAI
"x-ms-sk-completion-key" # Your API key

Running the starter

To run the console application within Visual Studio Code, just hit F5. As configured in launch.json and tasks.json, Visual Studio Code will run poetry install followed by python -m flask run sk_python_flask_chatgpt_plugin/app.py. Some users have had issues if there are spaces in their folder names.

To run the console application using CLI:

poetry install
python -m flask --app sk_python_flask_chatgpt_plugin/app.py run --port 5050 --debug

A POST endpoint exists at localhost:5050/skills/{skill_name}/functions/{function_name} For example, send a POST request to localhost:5050/skills/FunSkill/functions/Joke with the configuration headers and a JSON request body containing your prompt parameters such as: {"input": "time traveling to dinosaur age", "style": "wacky"}

For example,

curl -X POST \
  http://localhost:5050/skills/FunSkill/functions/Joke \
  -H 'Content-Type: application/json' \
  -H 'x-ms-sk-completion-model: text-davinci-003' \
  -H 'x-ms-sk-completion-endpoint: https://my-endpoint.openai.azure.com' \
  -H 'x-ms-sk-completion-backend: AZURE_OPENAI' \
  -H 'x-ms-sk-completion-key: Your API key' \
  -d '{"input": "time traveling to dinosaur age", "style": "wacky"}'

You may need to escape the double quotes if using curl in Windows

For simplicity, an endpoint also exists at localhost:5050/joke.

curl -X POST \
  http://localhost:5050/joke \
  -H 'Content-Type: application/json' \
  -H 'x-ms-sk-completion-model: text-davinci-003' \
  -H 'x-ms-sk-completion-endpoint: https://my-endpoint.openai.azure.com' \
  -H 'x-ms-sk-completion-backend: AZURE_OPENAI' \
  -H 'x-ms-sk-completion-key: Your API key' \
  -d '{"input": "time traveling to dinosaur age", "style": "wacky"}'

GorillaPlugin Functionality

The GorillaPlugin is integrated into the Flask application to process natural language commands and queue them for execution. It can either send the commands to a specified API endpoint or process them locally using the Gorilla CLI, depending on the presence of the api_endpoint query parameter in the request.

To use the GorillaPlugin, ensure that the GORILLA_CLI_PATH environment variable is set to the path of the Gorilla CLI executable on your system.

A POST endpoint exists at localhost:5050/gorilla/queue-commands where you can send a JSON payload with the command key containing the natural language command. Optionally, you can include the api_endpoint query parameter to specify an external API endpoint for processing the command.

For example:

curl -X POST \
  http://localhost:5050/gorilla/queue-commands \
  -H 'Content-Type: application/json' \
  -d '{"command": "list all files in the current directory"}'

If you want to use an external API endpoint:

curl -X POST \
  http://localhost:5050/gorilla/queue-commands?api_endpoint=http://external-api.com/process \
  -H 'Content-Type: application/json' \
  -d '{"command": "list all files in the current directory"}'

Using the starter as a ChatGPT plugin

First, run your Flask app locally. Then, follow instructions on the OpenAI website to install your plugin into ChatGPT. You may need to join a waitlist for developer access.

When installing into ChatGPT, use the domain localhost:5050 without http or https.

ChatGPT has an easier time deciding which route to use when you use a route per function, so the route localhost:5050/joke is the only one exposed to ChatGPT.

twilwa / sk-gorilla-cli-plugin

readme