Hey everyone, I'm Daksh, the developer behind Rexo Code. I started building Rexo because I wanted a terminal coding agent that wasn't tied to one AI provider. I also wanted to understand how these agents actually work under the hood, so I ended up building most of the system myself in Rust and spending a lot of time testing how it handles real coding tasks. Rexo can read a project, search through files, make changes, run shell commands, use MCP tools, ask for permission before sensitive actions, and continue working based on the results. During development I used Claude and Claude Code quite a lot for debugging, implementation and working through larger changes. I would test the results myself and decide what actually made sense for Rexo. v0.9 is the version I'm comfortable putting in front of more people. It's completely free and open source, and it currently supports Windows, Linux and macOS. I'm mainly here to get feedback from people who actually use coding agents. What would you change about the workflow? What features do you think are missing? And if you've built an agent yourself, I'd be especially interested in hearing how you approached the architecture.

Rexo Code

Provider-agnostic AI coding agent for your terminal

Provider-agnostic AI coding agent for your terminal