llumpy

Provider-agnostic client wrapper for prompting LLMs

🤖 Supported Providers

• Anthropic
• OpenAI
• Ollama

✨ Features

• One or Few Shot Prompting
• Streaming LLM Responses
• Load Prompts From File
• Asynchronous Variants
• Automatic Retry Handlers
• Custom Provider Clients

Installation

pip install git+https://github.com/dlg1206/llumpy.git@1.0.0

or add to requirements.txt

llumpy @ git+https://github.com/dlg1206/llumpy.git@1.0.0

Quickstart

1. Create a client

from llumpy.providers import ProviderFactory

# API key env var: ANTHROPIC_API_KEY
claude = ProviderFactory.anthropic('claude-sonnet-4-6')

# API key env var: OPENAI_API_KEY
gpt = ProviderFactory.openai('gpt-5.4')

# Ollama server url env var: OLLAMA_SERVER_URL (Default: http://localhost:11434)
llama3_latest = ProviderFactory.ollama('llama3')  # default ':latest'
llama3_8b = ProviderFactory.ollama('llama3', '8b')

Warning

Clients will fail to be initialized if API keys are bad, models do not exist, or key does not have access to that model.

2. Prompt

from llumpy.providers import ProviderFactory

llama3_8b = ProviderFactory.ollama('llama3', '8b')
response = llama3_8b.prompt_one("Hello!")
print(response)

See documentation for additional usage.