ChatGPT plugin that will (one day) allow us to manage cloud resources from ChatGPT

1. a local Flask web server that

• uses botocore to authenticate to AWS using a profile from .aws/config, given from the CLI when starting the web server
• serves a few static files required by the plugin documentation
• forwards requests received to AWS, authenticated with SigV4

1. the full EC2 OpenAPI YAML configuration taken from APIs.guru
2. a ChatGPT plugin configuration tying it all together

• install the requirements, preferably in a Python virtualenv

pip install -r requirements.txt

• start the web server:

./aws_proxy.py NAME_OF_PROFILE_FROM_AWS_CONFIG

• test the web server:

 curl http://localhost:3000/.well-known/ai-plugin.json
{
    "schema_version": "v1",
    "name_for_human": "EC2",
    "name_for_model": "gpt4",
    "description_for_human": "Manage your AWS resources.",
[...]

$ curl -X GET "http://localhost:3000/?Action=DescribeSecurityGroups&Version=2016-11-15"

<?xml version="1.0" encoding="UTF-8"?>
<DescribeSecurityGroupsResponse xmlns="http://ec2.amazonaws.com/doc/2016-11-15/">
    <requestId>22aba6e6-c226-487f-81d0-1cb0acc5aa04</requestId>
    <securityGroupInfo>
        <item>
[...]

When attempting to load it locally into ChatGPT the browser tab seems to hang, probably the EC2 API definition is too big and we need to expose less functionality.

After a while you'll get a bunch of errors in the ChatGPT plugin loader. This is expected and what we've got so far.

Any contributions are welcome, hack away and send us pull requests if you make some progress.

This software is licensed under the OSL-3 License.