GitHub
FreeTerminal AI Commander at your fingertips
FreeFree tier
Inputs: textOutputs: text, code
About GitHub
Instrukt is a terminal-based AI integrated environment that allows users to create, instruct, and manage modular AI agents. Built with Langchain, Textual, and Chroma, it offers a keyboard-and-mouse interface for power users. Agents are drop-in Python packages that can be extended, shared, and augmented with document indexes for retrieval-augmented generation (RAG). Tools can be attached or detached on-the-fly, and agents run inside secure Docker containers for sandboxed execution. The platform supports CUDA for GPU acceleration, remote access via SSH, and includes pre-built agents like a coding assistant and a QA chatbot. It is an open-source project under active development.
Key Features
Keyboard and mouse terminal interface with rich markdown output
Modular, drop-in Python agent packages that can be shared
Document indexing for question-answering with automatic language detection
Tool attachment/detachment to agents on-the-fly
Secure container execution via Docker
CUDA support for GPU acceleration
Remote access via SSH and terminal multiplexers
Pre-built agents (coding assistant, conversational QA)
Pros & Cons
Pros
- Fully open-source with a permissive license
- Modular agent design allows easy extension and sharing
- Document indexing enables powerful retrieval-augmented generation
- Sandboxed execution via Docker improves safety
- Terminal-based interface optimized for keyboard power users
- Supports multiple LLM backends through Langchain integration
Cons
- Work in progress; may contain bugs and API changes are expected
- Requires an OpenAI API key (may support others but not explicitly listed)
- Limited to terminal interface; no GUI or web UI
- Docker required for secure container features
Best For
Creating a coding assistant with RAG over a codebaseBuilding custom conversational agents for knowledge retrievalExperimenting with and testing modular AI agent architectures
FAQ
How do I install Instrukt?
Install via pip with `pip install instrukt[all]`, set your `OPENAI_API_KEY`, then run `instrukt`. Alternatively, clone the repository and install from source with poetry.
What are the system requirements?
Python 3, a terminal with 256-color support, and optionally Docker for container execution. For headless servers or containers, g++ and libmagic are needed.
Does Instrukt support GPU acceleration?
Yes, Instrukt supports CUDA for GPU acceleration and can run on bare metal or Docker containers with CUDA support.
Can I create my own agents?
Yes, agents are simple drop-in Python packages. You can design custom agents and tools, and share them with others.