Rigging
FreeLightweight LLM Interaction Framework (rust) 
FreeFree tier
Inputs: textOutputs: text
About Rigging
Rigging is a lightweight Python LLM framework built by dreadnode for using language models in production code. It features structured Pydantic model output interchangeable with unstructured text, support for hundreds of models via LiteLLM, prompts defined as Python functions with type hints and docstrings, simple tool use (even for models without native API support), connection-string style model/config management, integrated tracing with Logfire, chat templating, forking, continuations, async batching, and metadata/callback hooks. Designed for agents, chatbots, RAG pipelines, and security challenge solvers.
Key Features
Structured Pydantic model output interchangeable with unstructured text
LiteLLM as default generator supporting hundreds of models (OpenAI, Anthropic, vLLM, Hugging Face Transformers)
Prompts defined as Python functions with type hints and docstrings
Simple tool use even for models without native API tool support
Connection-string style model and config management (e.g., 'claude-3-sonnet-20240229')
Integrated tracing support with Logfire
Chat templating, forking, continuations, and generation parameter overloads
Async batching and iteration for large-scale generation
Metadata, callbacks, and data format conversions
Modern Python with type hints, async, Pydantic validation, and serialization
Pros & Cons
Pros
- Lightweight and flexible, easy to integrate into production Python code
- Supports a huge array of LLMs through a single interface (via LiteLLM)
- Native structured output with Pydantic models reduces parsing errors
- Async by default, enabling high throughput and batching
- Tool use works even with models that lack native tool APIs
- Open source and free to use
Cons
- Still a relatively new tool with a smaller community compared to established frameworks like LangChain
- Relies on LiteLLM as a dependency, which may introduce its own quirks
- Primarily Python-only; no official support for other programming languages
- Documentation and examples are still evolving
Best For
Building conversational chatbots and interactive agentsCreating Jupyter notebook code interpretersDeveloping agents for security challenges (e.g., OverTheWire, Damn Vulnerable Restaurant)Implementing Retrieval-Augmented Generation (RAG) pipelinesLarge-scale structured data generation with batching and async
FAQ
What is Rigging?
Rigging is a lightweight Python LLM framework by dreadnode that makes using language models in production code simple and effective. It supports structured Pydantic output, many models via LiteLLM, and async batching.
How do I install Rigging?
Install via pip: pip install rigging. You can also build from source using Poetry: cd rigging/ && poetry install.
Which LLMs does Rigging support?
Rigging supports any model from LiteLLM, vLLM, and Hugging Face Transformers. You pass an model ID in a connection string or set standard environment variables (e.g., OPENAI_API_KEY, ANTHROPIC_API_KEY).
Can Rigging output structured data?
Yes, Rigging supports structured Pydantic models interchangeably with unstructured text output. You can define prompts as Python functions with type hints and automatically parse responses into Pydantic models.