ChocoPichu Introduces Project Sulfur To Power Local AI Coding Agents
Project Sulfur is a new open source desktop application that lets you run AI coding agents entirely on your local machine. It allows the AI to read, write, edit, and search files in your workspace while keeping all data private. The program supports real-time streaming chat and saves your conversation history automatically.
Developer ChocoPichu created this tool to understand how local large language models operate at a fundamental level. They wanted to figure out how inference engines serve requests and handle streaming tool calls. By building this project over a few months, ChocoPichu produced a functional agent harness for others to use.
File management and hardware tuning
Key Features
- Runs completely local without any cloud.
- Supports multiple inference backends like llama.cpp.
- AI can read and edit your files.
- Exposed controls for GPU layers and threads.
- Ingests and analyzes PDF documents easily.
This application is for people who want to run AI models directly on their own hardware. Users can benefit from having deep controls over inference settings like context size and memory caching. It provides a practical way to manage workspaces and automate file tasks without sending data to external servers.
Project limitations and development notes
Currently the software only supports Windows, but the codebase uses pure Python so a Linux or Mac port is possible in the future. The developer notes this is their first open source project and was built primarily to learn rather than compete with commercial tools. It is not meant for production grade agentic setups yet.
"Built to learn, not to compete with Claude Code or Cursor if you need a production-grade agentic setup, this probably isn't it yet" - Source: Reddit
Project Links
