Analysis updated 2026-05-18
Run an AI coding assistant directly in your terminal to read, write, and edit code across your project.
Switch between different AI providers like Claude, GPT, or local Ollama models without changing tools.
Give the agent a specialist persona, such as a security auditor or backend architect, for focused help.
Keep a persistent project wiki and memory so the assistant does not relearn your codebase every session.
| bbarit/bbarit-agent-oss | cesarferreira/tmux.expose | episkey-g/groksearch-rs | |
|---|---|---|---|
| Stars | 41 | 41 | 40 |
| Language | Rust | Rust | Rust |
| Setup difficulty | easy | easy | moderate |
| Complexity | 2/5 | 2/5 | 3/5 |
| Audience | developer | developer | developer |
Figures from each repo's GitHub metadata at analysis time.
Ships as a single static binary with a one-line installer, requires an API key or key for at least one supported model provider.
bbarit-oss is a command line tool that acts as an AI coding assistant living right in your terminal. You type instructions in plain language and it reads your code, edits files, runs shell commands, and searches through your project to help you build and fix software, similar in spirit to tools like Claude Code or OpenAI's Codex CLI, but open source and self-hostable. It is written in Rust and ships as a single self-contained program with nothing else to install: no Node.js, no Python, no separate runtime. That means it starts up instantly and uses little memory, and the same one file works on macOS, Linux, and Windows. Installing it on Mac or Linux is one line in a terminal, and it can update itself in place afterward. One of its biggest selling points is flexibility in which AI model powers it. It connects to more than fifteen different AI providers and over a thousand models total, including Anthropic's Claude, OpenAI's GPT and Codex models, Google's Gemini, and local models run through Ollama, so you are not locked into one company's AI. The project also adds several features beyond a basic coding agent: a project wiki that stores what the AI has learned about your codebase so it does not have to relearn the architecture every session, a memory system that remembers facts about you and your project across sessions, nearly 300 curated personas that turn the assistant into a specialist such as a backend architect or security auditor, and the ability to reuse existing skills and connected tools already set up for Claude Code or Codex without reconfiguring them. The README explains that the project builds on the design of two other open source projects, an agent framework called Pi for its simple core loop, and qwen-code for its memory system, while the core Rust code itself was written from scratch. It documents this borrowing openly in a provenance file. This tool is aimed at developers who already work from a terminal and want an AI pair programmer they can run locally and control fully, rather than at non-technical users.
A self-hostable, open source AI coding assistant for the terminal, written in Rust, that works with over a thousand AI models from more than fifteen providers.
Mainly Rust. The stack also includes Rust, Anthropic Claude, OpenAI.
Use freely for any purpose, including commercial use, as long as you keep the copyright notice.
Setup difficulty is rated easy, with roughly 5min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.