whereismyai
Rax
Rax is an open-source framework for robot agents that scales online exploration and compiles programs.
RAX is an open-source framework for robot agents built around scaling exploration. A coding agent drives a robot through perception and control tools, learns tasks from its own online experience, and compiles successful decision sequences into modular programs. It supports single-arm, bimanual, dexterous, mobile, and humanoid robots in simulation and on real hardware. Supported backends include Isaac Sim, MuJoCo, Franka FR3, and Rainbow RB-Y1. The compiled programs replace the agent at runtime to execute tasks without demonstrations.
Rax at a glance
Key features
- Scale online exploration without demonstrations
- Compile exploration experience into executable code
- Support simulation and real hardware embodiments
- Maintain a skill library that grows across sessions
- Provide standardized MCP tools and agent skills
- Run in offline Docker containers or natively
Use it for
- Run simulation benchmarks for robot agents
- Control real robot arms and mobile manipulators
- Compile successful agent runs into fast programs
- Transfer learned task skills across sessions
Alternatives to Rax
- Rakazo — Complete ownership of data, models, and execution environment.
- Harness2 — It recursively refines agent execution structures using parallel and sequential search.
- VoltAgent — Offers open-source flexibility with built-in supervisor agent orchestration and RAG.
- OpenViking — Unifies agent memory, knowledge RAG, and skills into a single directory structure.
- AGiXT — Builds autonomous agents with integrated memory and web capabilities.
- OmniAgent — It evolves skills in real-time and dynamically hardens security during task execution.
- Composio — Equips AI agents with tools, managed user auth, and sandboxes across 1,500+ apps
- Atomic Agents — Build predictable, modular AI applications with consistent, reliable outputs.
- Pipecat — Builds real-time voice and multimodal agents with frame-level control.
- Webwright — Enables code-based web automation with reusable and rerunnable scripts.
- ARTEX — Runs autonomous multi-agent penetration testing with a dual-graph architecture.
- ART — Enhances LLM agent performance and reliability using reinforcement learning techniques.
All Rax alternatives