OmniAI::Anthropic

An Anthropic implementation of the OmniAI APIs.

Installation

gem install omniai-anthropic

Usage

Client

A client is setup as follows if ENV['ANTHROPIC_API_KEY'] exists:

client = OmniAI::Anthropic::Client.new

A client may also be passed the following options:

api_key (required - default is ENV['ANTHROPIC_API_KEY'])
host (optional)

Configuration

Global configuration is supported for the following options:

OmniAI::Anthropic.configure do |config|
  config.api_key = '...' # default: ENV['ANTHROPIC_API_KEY']
  config.host = '...' # default: 'https://api.anthropic.com'
end

Chat

A chat completion is generated by passing in prompts using any a variety of formats:

completion = client.chat('Tell me a joke!')
completion.text # 'Why did the chicken cross the road? To get to the other side.'

completion = client.chat do |prompt|
  prompt.system('You are a helpful assistant.')
  prompt.user('What is the capital of Canada?')
end
completion.text # 'The capital of Canada is Ottawa.'

Model

model takes an optional string (default is claude-3-haiku-20240307):

completion = client.chat('Provide code for fibonacci', model: OmniAI::Anthropic::Chat::Model::CLAUDE_SONNET)
completion.text # 'def fibonacci(n)...end'

Anthropic API Reference model

Temperature

temperature takes an optional float between 0.0 and 1.0 (defaults is 0.7):

completion = client.chat('Pick a number between 1 and 5', temperature: 1.0)
completion.text # '3'

Anthropic API Reference temperature

Stream

stream takes an optional a proc to stream responses in real-time chunks instead of waiting for a complete response:

stream = proc do |chunk|
  print(chunk.text) # 'Better', 'three', 'hours', ...
end
client.chat('Be poetic.', stream:)

Anthropic API Reference stream

Format

format takes an optional symbol (:json, :text, or OmniAI::Schema::Format) and modifies the system message to include additional context for formatting:

format = OmniAI::Schema.format(name: "Contact", schema: OmniAI::Schema.object(
  properties: {
    name: OmniAI::Schema.string,
  },
  required: %i[name]
))
completion = client.chat(format: :json) do |prompt|
  prompt.user('What is the name of the drummer for the Beatles?')
end
format.parse(completion.text) # { "name": "Ringo" }

Anthropic API Reference control-output-format

Computers

sudo apt-get install convert # screenshots
sudo apt-get install scrot # screenshots
sudo apt-get install xdotool # mouse / keyboard

computer = OmniAI::Anthropic::Computer.new

completion = client.chat(tools: [computer]) do |prompt|
  prompt.user('Please signup for reddit')
end

Extended Thinking

Extended thinking allows Claude to show its reasoning process. This is useful for complex problems where you want to see the model's thought process.

# Enable with default budget (10,000 tokens)
response = client.chat("What is 25 * 25?", model: "claude-sonnet-4-20250514", thinking: true)

# Or specify a custom budget
response = client.chat("Solve this complex problem...", model: "claude-sonnet-4-20250514", thinking: { budget_tokens: 20_000 })

When thinking is enabled:

Temperature is automatically set to 1 (required by Anthropic)
max_tokens is automatically adjusted to be greater than budget_tokens

Accessing Thinking Content

response.choices.first.message.contents.each do |content|
  case content
  when OmniAI::Chat::Thinking
    puts "Thinking: #{content.thinking}"
    puts "Signature: #{content.metadata[:signature]}" # Anthropic includes a signature
  when OmniAI::Chat::Text
    puts "Response: #{content.text}"
  end
end

Streaming with Thinking

client.chat("What are the prime factors of 1234567?", model: "claude-sonnet-4-20250514", thinking: true, stream: $stdout)

The thinking content will stream first, followed by the response.

Anthropic API Reference thinking