# Reflex

> Stream observations in, get actions out. Low-latency infrastructure for AI-native robotics.

## Docs

- [rfx deploy](https://docs.tryreflex.ai/cli/deploy.md): Load a trained policy and run it on hardware.
- [rfx doctor](https://docs.tryreflex.ai/cli/doctor.md): Check your environment before running other commands.
- [CLI overview](https://docs.tryreflex.ai/cli/overview.md): Three primary commands. Small on purpose.
- [rfx record](https://docs.tryreflex.ai/cli/record.md): Collect robot observations into a LeRobot dataset.
- [Architecture](https://docs.tryreflex.ai/concepts/architecture.md): A compact core for modern AI robotics workloads where simulation, collection, policy execution, and real hardware share the same runtime contracts.
- [Artifacts](https://docs.tryreflex.ai/concepts/artifacts.md): Self-describing policy directories. Save once, load anywhere.
- [Core primitives](https://docs.tryreflex.ai/concepts/primitives.md): The five package surfaces that make up the Reflex SDK.
- [Robot interface](https://docs.tryreflex.ai/concepts/robot-interface.md): Three methods. The same contract for simulation, mock, and real hardware.
- [Collection](https://docs.tryreflex.ai/guides/collection.md): Recording observations into structured datasets is a first-class primitive.
- [Deploy](https://docs.tryreflex.ai/guides/deploy.md): Load a trained policy and run the control loop. One command.
- [Operator checklist](https://docs.tryreflex.ai/guides/operator-checklist.md): Preflight gates to run before every production deploy.
- [Simulation](https://docs.tryreflex.ai/guides/simulation.md): Run rfx without robot hardware. Genesis is available today, MuJoCo and Isaac Lab are next.
- [SO-101 quickstart](https://docs.tryreflex.ai/guides/so101.md): Fastest path to a working demo on a physical SO-101 arm.
- [Installation](https://docs.tryreflex.ai/installation.md): Install the rfx SDK and verify the runtime is ready.
- [Introduction](https://docs.tryreflex.ai/introduction.md): Stream observations in, get actions out. Reflex is a low-latency infrastructure layer for AI-native robotics.
- [Quickstart](https://docs.tryreflex.ai/quickstart.md): Install rfx, record a demo, and deploy a policy.
- [Deploy API](https://docs.tryreflex.ai/sdk/deploy.md): rfx.deploy() — load a policy and run the control loop.
- [Hardware configs](https://docs.tryreflex.ai/sdk/hardware-configs.md): Built-in robot configurations.
- [Model management](https://docs.tryreflex.ai/sdk/model-management.md): Save, load, inspect, and push policies.
- [MotorCommands](https://docs.tryreflex.ai/sdk/motor-commands.md): Named joint control. No hardcoded tensor indices.
- [Python SDK](https://docs.tryreflex.ai/sdk/overview.md): One Python API for the full AI robotics loop.
- [Policies](https://docs.tryreflex.ai/sdk/policies.md): A policy is any callable Dict[str, Tensor] -> Tensor.

## OpenAPI Specs

- [openapi](https://docs.tryreflex.ai/api-reference/openapi.json)