Skip to content

muaz

A local, web-first LLM agent platform. Chat with agents, run multi-agent pipelines, and wire in any OpenAI-compatible model, Google Gemini, local Ollama, or AWS Bedrock — from a single binary that runs on your own machine.

Single binary

One binary. Bare muaz opens the browser UI, muaz chat the terminal REPL — no runtime, no containers, no setup beyond an API key.

Any model

OpenAI-compatible endpoints, Google Gemini, local Ollama, or AWS Bedrock. Switch providers per agent with a few lines of YAML.

Multi-agent pipelines

Wire agents together in sequential, parallel, or router pipelines. Run them from the REPL with /run or from the shell with muaz pipeline run.

Runs on your machine

The UI is served by a local web server on 127.0.0.1. Your prompts, files, and sessions never leave your computer — no cloud required.

Terminal window
# macOS / Linux
curl --proto '=https' --tlsv1.2 -LsSf \
https://github.com/shigar-dev/releases-muaz/releases/latest/download/muaz-installer.sh | sh
export OPENAI_API_KEY=sk-...
muaz

On first run, muaz bootstraps ~/.muaz/ with a default agent and sensible defaults. Head to Installation for all platforms and options.