Hax

This tool is a terminal native coding agent written in C. It ships as a single native binary with a small dependency set, starts instantly and uses only a few megabytes of memory, which leaves more RAM free for local models. It runs on Linux, macOS, FreeBSD and OpenBSD, and on Windows under WSL. It is aimed at developers who live in the terminal, run local models, want to audit what their tools do, or work where resources are scarce. Local models are first class: it auto discovers the model and runtime capabilities, so a llama.cpp server works without a custom provider block. It also supports OpenAI and compatible endpoints, Anthropic and compatible endpoints, Codex through a ChatGPT subscription, OpenRouter and OpenCode Zen and Go. Streaming Markdown and live tool output are reflowed for the terminal while native scrollback is preserved, and a transcript view shows exactly what was sent to the model and what came back, with an optional detailed wire protocol trace. It behaves like a Unix tool, with XDG paths, clean stdout in one shot mode, plain text config and session files, and composition through subprocesses rather than plugins. It deliberately leaves out MCP marketplaces, a plugin runtime, IDE panels and per command permission prompts, and its documentation explains each omission. Install it with Homebrew, from the AUR on Arch Linux, as a prebuilt static binary for x86_64 or aarch64, or by building from source. No pricing is published on the site; the project is distributed through public repositories and you bring your own model provider.