New to the ExecutionPlatform?

It is a Processor-in-the-Loop test system: your microcontroller sits on a plug-in board, and the test system drives and observes every pin.

Read the overview first

LLMs are blind to physics

An AI coding agent writes driver and HAL code in minutes. Whether that code sets the right pin at the right time, it cannot tell. The ExecutionPlatform tests exactly this boundary, the transition from firmware to electrical signal at the microcontroller pin.

All input is text: Robot Framework, Python and YAML.

All actions are automatic. No GUI, no manual step.

People and agents use the same interface.

The agent closes the loop on real hardware

Writes the test, the firmware glue and the ep-config.

Builds the ep-config.

Runs the test. The ExecutionPlatform drives and measures the DUT pins.

Reads the result and fixes the code until the test passes.

Shipped with the software package

The software package contains everything an agent needs to write and run tests without guessing.

AGENTS.md, read automatically by Claude Code, Cursor, GitHub Copilot and Codex.

A machine-readable list of all keywords and model functions per endpoint.

A command line tool that validates and builds ep-configs, for agents and CI alike.

AGENTS.md tells the agent what it needs to act on its own:

Which test mode fits the goal: unit, integration or system test.

How names in ep-configs, tests and models connect.

Error messages with their cause and fix.

Read the software package documentation

You stay in control

AI support covers Robot Framework today.
pytest follows in the next release.

The ExecutionPlatform contains no AI.

Your agent works in your repository and with your tools.

Questions about AI agents

Steps we recommend to learn more about the ExecutionPlatform.

Get started