QA-Pilot

Free

An interactive chat project that leverages Ollama/OpenAI/MistralAI LLMs for rapid understanding and navigation of GitHub code repository or compressed file resources.

FreeFree tier

Outputs: text

Type

Open Source

Visit Website GitHub (38 stars)

About QA-Pilot

QA-Pilot is an open-source interactive chat project that allows users to rapidly understand and navigate GitHub code repositories by cloning them and engaging in natural language conversations. It supports a wide variety of local and cloud-based LLM providers including Ollama, OpenAI, MistralAI, LocalAI, ZhipuAI, Anthropic, LLaMA.cpp, NVIDIA, Tongyi, and Moonshot. The tool features chat history storage, multiple concurrent chat sessions, a session search function, and integration with CodeGraph for visualizing Python code. Built on FastAPI with a Svelte frontend, it offers easy configuration via a config file and is designed for local or private deployments. The project is explicitly a test project and not intended for production use or analysis of sensitive data.

Key Features

Chat with GitHub public repositories via git clone

Store chat history and manage multiple chat sessions

Search function to quickly locate previous sessions

Integration with CodeGraph to view and explore Python files

Support for a wide range of LLM providers: Ollama, OpenAI, MistralAI, LocalAI, ZhipuAI, Anthropic, LLaMA.cpp, NVIDIA, Tongyi, Moonshot

Easy configuration via config/config.ini

FastAPI backend with Svelte frontend for improved performance

Option to use FlashRank for search ranking

Upload model function for LLaMA.cpp

Prompt templates setting support

Pros & Cons

Pros

Supports a wide range of LLM providers (open-source and commercial)
Free and open-source with local execution options for data privacy
FastAPI backend with Svelte frontend for good performance
Chat history and session management with search functionality
CodeGraph integration for Python code visualization
Easy configuration and support for multiple model uploads

Cons

Project is explicitly a test project and not production-ready
Requires git clone of each repository, which may be slow for large codebases
Manual setup of providers, dependencies, and virtual environment needed
No built-in authentication or multi-user support
Primarily designed for public GitHub repositories; private repos require additional configuration

Best For

Rapidly understand and navigate unfamiliar GitHub codebasesChat with code repositories for documentation, learning, or code reviewPerform local code analysis using offline LLMs (e.g., Ollama, LLaMA.cpp) for data privacyExplore code structure and dependencies through natural language queries

FAQ

What LLM providers are supported by QA-Pilot?

QA-Pilot supports Ollama, OpenAI, MistralAI, LocalAI, ZhipuAI, Anthropic, LLaMA.cpp, NVIDIA, Tongyi, and Moonshot.

Is QA-Pilot production-ready?

No. The project is explicitly a test project to validate the feasibility of a fully local solution for question answering using LLMs and vector embeddings. It is not meant for production use or for analyzing critical, customer, or private/sensitive data.

How do I install QA-Pilot?

Clone the repository, create a conda environment with Python 3.10.14, install dependencies from requirements.txt, install PyTorch with CUDA, and set up the desired LLM provider (e.g., Ollama, LocalAI). Full instructions are in the repository README.