Hax is a minimalist, native CLI coding agent created by Oleksandr Chekhovskyi. Written in C and designed for POSIX-compliant environments, it provides terminal-based automated development workflows with an emphasis on low overhead, speed, and standard Unix behavior across Linux, macOS, FreeBSD, and OpenBSD. In practical usage, Hax operates from the shell to orchestrate multi-step code generation, file editing, and command execution tasks. It gathers contextual instructions from repository files such as AGENTS.md and integrates with standard LLM backends—including local inference servers like llama-server—to execute tool calls, run shell subcommands, and track session state. What sets Hax apart from other coding agents is its hyper-lightweight system architecture. While many agents require multi-megabyte Node.js or Python runtimes that consume significant memory, Hax is optimized for tiny private memory footprints and minimal execution latency. It also adheres to cross-agent standards by discovering tools in .agents/skills and standard XDG configuration paths. Hax is an open-source project hosted on GitHub and available under an open-source license.
Native C architecture — Delivers near-instant startup times and an ultra-low private memory footprint compared to runtime-heavy agents. - Broad POSIX and BSD support — Fully supports Linux, macOS, FreeBSD, and OpenBSD, incorporating OS-specific kernel accounting and resource metrics. - Standardized skill discovery — Automatically discovers modular capabilities from project-level .agents/skills directories up to repository roots, as well as ~/.agents/skills and XDG paths. - AGENTS.md context ingestion — Discovers and reads AGENTS.md files throughout repository directory trees to maintain project context. - Session resumption — Tracks interaction history and reproducible state ordering with native session continuation using --continue. - Subprocess and resource tracking — Efficiently runs and monitors subprocesses for shell execution and code manipulation while accounting for tool processes.
Use cases
Use Case: Resource-Constrained and Embedded Development Developers running coding agents on low-memory servers, single-board computers, or alongside local quantized models via llama-server can run Hax without competing for RAM or CPU overhead.
Use Case: BSD and Alternative Unix Workstations Engineers on FreeBSD or OpenBSD environments can utilize a fully supported, native AI coding agent where Python/Node-based alternatives often struggle with platform compatibility.
Use Case: Standardized Skill-Based Tooling Teams that maintain shared cross-agent tools in .agents/skills can deploy Hax across repositories without writing custom harness adapters or rewriting agent instructions.
Who it is for
Hax is targeted at systems developers, Unix enthusiasts, and command-line purists who require a fast, low-overhead agent that runs across diverse POSIX operating systems. It is less suitable for developers who need bundled IDE graphical extensions, web UI dashboards, or complex cloud-hosted agent harnesses.
Install guide
Installation Clone the repository and run the dependency installation script for your operating system: bash git clone https://github.com/OleksandrChekhovskyi/hax.git cd hax ./install_deps.sh make
First Run Configure your environment with your model provider API key or point Hax to a local inference server: bash export OPENAI_API_KEY="your-api-key-here" # Or configure a local endpoint if running llama-server Ensure project-specific context is placed in AGENTS.md or skill scripts are inside .agents/skills/.
Example Session Run Hax directly from the root of your project directory: bash # Start a new task hax "Refactor the HTTP client in src/network.c to handle connection timeouts" # Resume the last interactive session hax --continue
Gemini CLI — Gemini CLI is an open-source terminal coding agent maintained under the Google Gemini organization (google-gemini/gemini-cli). It provides a command-line and…
Open Interpreter — Open Interpreter is an open-source terminal coding agent maintained by Killian Lucas and the Open Interpreter community. It is designed…
Crush — Crush is an open-source terminal coding agent created by Charm (Charmbracelet), the team behind popular Go terminal UI libraries like…
Codel — Codel is an open-source, self-hosted autonomous software engineering agent created by semanser. It combines an execution backend, an interactive web…
San — San is an open-source terminal coding agent developed by the genai-io organization on GitHub. Built natively in Go, San is…
amux — amux is an open-source agent harness and session multiplexer developed by Mixpeek. Built in Rust, it is designed to manage,…