This gem implements autoflux agent to use OpenAI as the backend.
Install the gem and add to the application's Gemfile by executing:
bundle add autoflux-openaiIf bundler is not being used to manage dependencies, install the gem by executing:
gem install autoflux-openaiThe use the agent, set the OPENAI_API_KEY environment variable that the agent can use to authenticate with OpenAI.
agent = Autoflux::OpenAI::Agent.new(model: "gpt-4o-mini")
res = agent.call("Hello, world!")
# => "Hello, world!" from OpenAIThe agent will return
contentfrom the response as the interface of theautofluxagent.
If you want to use as multi agents the agent name can be set.
agent = Autoflux::OpenAI::Agent.new(name: "shopping", model: "gpt-4o-mini")
res = agent.call("Hello, world!")
# => "Hello, world!" from OpenAIThe agent use same parameters to call the OpenAI API. You can customize it when creating the agent.
agent = Autoflux::OpenAI::Agent.new(
model: "gpt-4o-mini", # Required
# Extra parameters
temperature: 0.5,
)
res = agent.call("Hello, world!")
# => "Hello, world!" from OpenAIThe
model,messages, andtoolswill be override by the agent. The stream mode is not supported.
You can attach tool to the agent to give it more capabilities.
uppercase = Autoflux::OpenAI::Tool.new(
name: "uppercase",
description: "Convert the content to uppercase",
parameters: {
type: "object",
properties: {
text: {
type: "string"
}
}
}
) do |params|
{ text: params[:text].upcase }
end
agent = Autoflux::OpenAI::Agent.new(
model: "gpt-4o-mini",
tools: [uppercase],
memory: [
{ role: "system", content: "Always transform the user input and don't do anything else." }
]
)
res = agent.call("Hello, world!")
# => "HELLO, WORLD!" from OpenAIThe tool is an object has a name, description, parameters, and a #call method. The #call method will be called with the parameters from the agent.
class MyTool
attr_reader :name, :description, :parameters
def initialize
@name = "my-tool" # Must be string which can be lookup by the agent
@description = "My tool"
@parameters = {
type: "object",
properties: {
text: {
type: "string"
}
}
}
end
def call(params, **context)
{ text: params[:text].upcase }
end
endThis gem embeds a lightweight client to interact with OpenAI API. The client can be used to interact with OpenAI API directly.
client = Autoflux::OpenAI::Client.new(api_key: "your-api-key")
res = client.call(
model: "gpt-4o-mini",
messages: [{ role: "user", content: "Hello, world!" }]
)
# => { choices: [{ message: { role: "assistant", content: "Hello World!" }}] }If your api key or endpoint is not default, you can specify them in the client.
client = Autoflux::OpenAI::Client.new(
api_key: "your-api-key",
endpoint: URI("https://gateway.ai.cloudflare.com/v1/{account_id}/{gateway_id}/openai")
)
client.call(
model: "gpt-4o-mini",
messages: [{ role: "user", content: "Hello, world!" }]
)The agent default use a Ruby array to store the conversation history. If you want to use a different memory store, you can implement the Autoflux::OpenAI::_Memory interface and pass it to the agent.
class MyMemory
def initialize
@store = []
end
def push(message)
@store.push(message)
end
def <<(message)
push(message)
end
def to_a
@store.last(100)
end
endAfter checking out the repo, run bin/setup to install dependencies. Then, run rake spec to run the tests. You can also run bin/console for an interactive prompt that will allow you to experiment.
To install this gem onto your local machine, run bundle exec rake install. To release a new version, update the version number in version.rb, and then run bundle exec rake release, which will create a git tag for the version, push git commits and the created tag, and push the .gem file to rubygems.org.
Bug reports and pull requests are welcome on GitHub at https://github.com/elct9620/autoflux-openai.
The gem is available as open source under the terms of the Apache License 2.0.